15:45   Zuiderduinzaal: Poster session
H. Verhoeven, Pauly Ossenblok, S. van Liempt, E. Vermetten, P. Boon, Klaas Nicolay
Abstract: Sleep spindles are a characteristic feature of NREM sleep, present especially in sleep stage 2. These spindles are believed to be relevant for the processing of daily experience and are suggested to play a role in memory consolidation, especially the fast spindles with frequency 13-16 Hz [1]. Event related correlation of EEG and fMRI is used to study the underlying neuronal network of these spindles [2], which might reveal alteration in brain activity during sleep spindles in veterans suffering from posttraumatic stress disorder (PTSD), who commonly have sleep complaints and decreased memory performance. Simultaneous EEG and fMRI were recorded during 45 minutes. The subjects were lying in a supine position and were encouraged to fall asleep. Therefore, the recordings were performed at 11 or 12 o’clock in the evening. For the EEG, offline artefact removal reduced gradient and cardioballistic artefacts, thus allowing the visual identification of distinct sleep stages and spindles. In order to differentiate between fast and slow spindles (11-13 Hz) the continuous wavelet transform was performed using a Morlet mother wavelet. The wavelet transform was performed at an epoch of 2.5 sec preceding and following the scored spindle onset, in a frequency range of 9-18Hz. Next, the wavelet coefficient spectrum was calculated from the correlation coefficients [3]. Peaks in the spectrum allowed identification of the dominant frequencies of the spindle. The onset time of the spindle was determined by a 95% threshold of the difference in energy of the correlation coefficients of the spindle compared to the background signal. A paradigm for a general linear model (GLM) was derived from the spindles occurrence in the EEG, for the analysis of the fMRI data. The time-frequency analysis used in this method, the continuous wavelet transform using a Morlet mother wavelet, is suitable for distinguishing fast spindles from background EEG and from the slow spindles. Furthermore, determination of the onset of the spindle was more accurate than identification on the basis of visual review, thus allowing an unambiguous correlation between the spindle occurring in the EEG and fMRI. Since timing of the events is of major importance in event related correlation of EEG and fMRI, and spindle onset times can be determined more accurate than by visual scoring, the described method enabled better correlation of EEG and fMRI with the intention of unravelling the neuronal networks related to spindles. These neuronal networks might reveal information about decreased sleep dependent memory consolidation in veterans suffering from PTSD.
Stefan Arsov, Reindert Graaff, Gerard Rakhorst, Rutger Ploeg
Abstract: Objectives The aim of this study is the measurement of the accumulation of Advanced Glycation End-products (AGEs) in the skin of Hemodialysis (HD) patients waiting for kidney transplantation and to assess the prognostic value of the AGE measurements on cardiovascular disease (CVD) morbidity and mortality. Methods The accumulation of AGEs was measured using a Skin Autofluorescence Reader (AFR) in 170 end stage renal disease patients. Patients were divided in Diabetes Mellitus (DM) + and -. Effects of gender, age, dialysis duration, smoking and HCV was investigated. The AF values were defined as the ratio between the fluorescenting and the reflecting light. [1] Beside the skin AF as potential marker of CVD mortality and morbidity, routine biochemical analyses and dialysis parameters were examined. Multiple regression statistics was used for determination of the prognostic value of the markers. Morbidity and mortality will be recorded over a period of 3 yrs. Results First measurements showed that AGEs and the prevalence of CVD are higher (p<0,01) in the DM+ patients (3,16±0,93) then in the DM- patients (2,73±0,86). The level of AGE’s and the CVD prevalence correlate with the age (r=0,16, p<0,05) of the patients and the time they were on dialysis (r=0,17, p< 0,05). In addition we found higher values (p<0,05) of AGEs in patients that were HCV + (2,92±0,88 vs 2,62±0,89), which represent 116 (68%) of the population. In the multiple regression model for the total HD population the CRP(p<0,05), age (p<0,01) and DM+ (p<0,01) were the strongest predictive markers ,while in the DM groups the AGEs were the best (p<0,01). Conclusions The AGEs concentration in HD patients is associated with DM+, age and the duration of dialysis. The concentration of AGEs (p<0,01) measured in the skin in relationship with the DM was the best CVD prediction marker. [1]Meerwaldt, R. and Smit ,A.J.. "Skin autofluorescence, a measure of cumulative metabolic stress and advanced glycation end products, predicts mortality in hemodialysis patients." J.Am.Soc.Nephrol. 16.12 : 3687-93. ,2005
Reza Lotfi, Wouter Serdijn
Abstract: In all biomedical sensor-interface ICs or data acquisition systems, the employed analog-to-digital converter (ADC) plays an important role in the entire system accuracy and power consumption. Successive-approximation analog-to-digital converters (SA-ADCs) have recently been widely used for moderate-speed moderate-resolution biomedical applications where the power consumption is the major concern including implantable pacemakers or hearing-aid devices [1-4]. In this work, several techniques are employed to further reduce the power consumption of a SA-ADC with specifications suitable for most implantable or wearable biomedical applications. These solutions include a split-segmented architecture for the capacitor-based digital-to-analog converter (DAC) employed in the SA-ADC. In this DAC, unary-weighted capacitors controlled by more-significant bits, already converted to thermometer code, are split from binary-weighted capacitors controlled by the less-significant bits by a splitting capacitors leading to less power consumption. Also, a modified switching scheme for the DAC is proposed where the most-significant bit is not reset at the beginning of conversion of each new sample. Furthermore, by employing a smaller supply voltage for the comparator and the successive-approximation register while using a new power-efficient digital level converter more power is saved. Spectre simulation results of a single-ended 10-bit 100kS/s SA-ADC in a 0.13-µm CMOS technology employing the proposed techniques show that the ADC (excluding reference buffers) consumes less than 1 µW of power while the effective number of bits is 9.1.
Yvonne Schroeder, Jacques Huyghe, Rene van Donkelaar, Keita Ito
Abstract: Back pain is a frequently occurring complaint in adults, having a relatively large impact on the European economy as it often partially incapacitates the patient (1). Intervertebral discs are believed to be a key element of back pain. FE models have become an important tool to study load distribution in the healthy and degenerated disc. The constitutive law rather than the geometric features are an essential issue in modeling the intervertebral disc (2). However in most models, degeneration is simulated phenomenologically, but not according to matrix composition. We have developed a novel 3D disc model that accounts for prestressing of collagen fibers due to physical principles of osmosis and is based on experimentally quantified material properties from human tissue that are dependent on matrix biochemical composition (3) . In this study, the 3D disc model is corroborated with experimental data of whole discs from literature (4;5). The bulging and creep behavior of the 3D disc model was confronted with experiments of whole discs from the literature (radial bulging, height change and intradiscal pressure). The 3D FE model resembles 1/4 of a full disc with nucleus and annulus and describes each tissue as a biphasic material, consisting of a porous solid matrix saturated with water. The solid matrix is further divided into a non-fibrillar and fibrillar matrix (3;6). During corroboration 4 adaptations to the model were explored. The fibrillar matrix was extended to incorporate smaller secondary fibril structures, e.g. minor collagen, elastin, etc. The shear stiffness was also varied. Radial bulge, axial creep deformation and intradiscal pressure were in good agreement with the experiments from the literature, provided that the simplified fiber structure was extended with a more complex secondary fiber structure. This reduced the deformability of the 3D model and a slightly increased shear stiffness ensured convergence of the highly non-linear simulations. The evaluated 3D disc model may now be used to explore the biomechanical implications of disc degeneration on its function and integrity as well as to explore therapeutic mechanisms for repair and regeneration. REFERENCES (1) Op De Beeck R., Hermans D. Research on work-related low back disorders. European Agency for Safety and Health at work; 2000. (2) Bowden A. Finite Element Modeling of the Spine. In: Kurtz SM, Edidin AA, editors. Spine Technology Handbook. Elsevier Academic Press; 2006. p. 443-72. (3) Schroeder Y, Wilson W, Huyghe JM, Baaijens FP. Osmoviscoelastic finite element model of the intervertebral disc. Eur Spine J 2006 Aug;15 Suppl 15:361-71. (4) Brinckmann P, Horst M. The influence of vertebral body fracture, intradiscal injection and partial discectomy on the radial bulge and height of human lumbar disks. Journal of Biomechanics 1985;18(7):546. (5) Heuer F, Schmitt H, Schmidt H, Claes L, Wilke HJ. Creep associated changes in intervertebral disc bulging obtained with a laser scanning device. Clin Biomech (Bristol , Avon ) 2007 Aug;22(7):737-44. (6) Schroeder Y, Elliott DM, Wilson W, Baaijens FP, Huyghe JM. Experimental and model determination of human intervertebral disc osmoviscoelasticty. J Orthop Res 2008 Aug; 26(8):1141-6.
Herke Jan Noordmans, Sander Velthorst, Henk van den Brink, Joris Jaspers, Ruud Verdaasdonk
Abstract: To assure the optical quality of rigid endoscopes in clinical practice, a test bench has been developed to measure the illumination pathway using a LED and photo cell and the viewing pathway using a LCD generated test pattern and high resolution camera. By storing the results into a database, changes in quality can be followed over time. In this paper the results are presented of four months of measurement at the central sterilization department after cleaning and before sterilization. In total 355 measurements were performed on 119 rigid endoscopes of 28 different types. The accuracy of the measurements with the test bench proved to be within 5% for the illumination fibers and within 20% for the image contrast test which is ascribed to the intensity fluctuations of the LCD screen. The clinically used endoscopes appeared to degrade slowly and their lifetime, based on reaching 20% of their initial quality, was estimated at 2 to 3 years. As expected brand new endoscopes have a better quality compared to clinically used endoscopes of the same type. However, one new endoscope performed less and was returned to the manufacturer for inspection. Endoscopes returned from repair were better than used endoscopes but could not be restored to the quality of new endoscopes. Remarkably, the quality of repaired endoscopes dropped quickly within the first months after clinical use. The use of a test bench to monitor the optical quality of endoscopes over time shows to be a valuable investment that will contribute to the quality of patient treatment. More research is required to reveal the cause of degradation and improve the significance and accuracy of the measurements. REFERENCES [1] H.J. Noordmans, S. Kruit, P. Stroosnijder, H. vd Brink, R.M. Verdaasdonk, “Quantitative assessment of degradation of the optical quality of rigid endoscopes in clinical practice”, Proc. of SPIE Vol. 6849, 68490K, (2008). [2] Rapportage van de projectgroep MICADO, “Kwaliteitsborging van Instrumenten en Apparatuur gebruikt bij Minimaal Invasieve Chirurgie”, June 2008, www.wibaz.nl.
Danijela Vukadinovic, Theo van Walsum, Rashindra Manniesing, Sietske Rozie, Aad van der Lugt, Wiro Niessen
Abstract: There is large interest in automated carotid artery plaque analysis, as there is increasing evidence that atherosclerotic plaque volume and composition are related to acute cerebrovascular events. In this study, a novel semi-automatic method was developed to segment the plaque in the common carotid artery. First the carotid lumen is semi-automatically segmented using a level set approach initialized with three seed points [1]. Then calcified regions located within the vessel are automatically detected with the GentleBoost framework using a number of descriptive features. In the next step, voxels are classified as being inside or outside vessel with the GentleBoost framework with a similar set of features [2]. Finally, a 2D ellipse shape deformable model is fitted to the combined result of the calcium and vessel wall classification. The ellipses are fitted in a way that automatically detected calcified regions are positioned inside each ellipse, at the same time the sum of intensities along the ellipse is minimized. The method was trained on 40 CTA datasets, and tested on 60 data sets. For 20 datasets from the test set, second observer manual segmentations were available. The evaluation included comparison between the automated and manual segmentation and a comparison with interobserver variability. Automated segmentation revealed a vessel volume of 897±722mm3, while difference between manual and automated segmentation was 3.1±116 mm3 (p=0.92), the similarity index (SI) was 91% and the coefficient of variation (CoV) is 13%. This was better than the interobserver variability with a difference of -20±118 mm3 (p=0.46), an SI of 92% and a CoV of 15%. Automated segmentation revealed a plaque volume (vessel volume–lumen volume) of 393±295 mm3, while difference between manual and automated segmentation was -4.7±116 mm3 (p=0.75), the SI was 78% and the CoV is 29%. This was better than the interobserver variability with a difference of -37±121 mm3 (p=0.19), an SI of 80% and a CoV of 33%. We conclude that automated segmentation of the atherosclerotic plaque is robust and accurate when compared to the manual interobserver variability. REFERENCES [1] Manniesing R, Viergever MA, Niessen WJ. “Vessel Axis Tracking Using Topology Constrained Surface Evolution,” IEEE Transaction on Medical Imaging, 26(3), 309-316 (2007) [2] J.Friedman, T.Hastie, R. Tibishirani, “Additive Logistic Regression: A Statistical View of Boosting”, The Annals of Statistics, vol. 28, pp.337-407, 2000
Michiel Grashuis
Abstract: A structured design methodology for on-chip, analog wavelet filters for biomedical signal analysis is proposed. In many wearable and implantable devices, the analog-to-digital converter (ADC) takes up a large chunk of the device’s power consumption. For this reason, we propose to reduce the overall power consumption by performing analog pre-processing employing a wavelet filter. Despite consuming power itself, the wavelet filter allows for a reduction of the ADC resolution and hence may entail a favorable reduction in power consumption of the entire system. A wavelet filter is a device which has a (time-reversed) mother wavelet as its impulse response and implements one scale of the corresponding wavelet transform. We present a design flow that can generate a wavelet filter topology from an arbitrary given wavelet base [1]. The design flow is implemented in Matlab code. It requires a function description as input and produces an optimized state-space description and wavelet filter topology. The program first requires the user to define the wavelet base in the form of a function description (in the time domain). This can be any known wavelet, e.g., a Gauss1 (1st derivative of a Gaussian), Daubechies or Morlet wavelet, or a (multi-) wavelet tailored to the detection of a particular signal morphology [2]. The user of the program then has the option to choose between a continuous-time (CT) and a discrete-time (DT) system. For the DT wavelet filter, the user subsequently can define the sampling frequency (Fs) and the nature of the basic building block employed, e.g. a delay (1/z) element or an integrator (1/(z-1)) element. Based on the proper choice of this element, the system can be consequently implemented by means of a switched current (SI) or switched capacitor (SC) circuit technique. For the CT case, the only active element the designer can select is the integrator (1/s), which can be implemented, e.g., using gm-C, opamp-RC or dynamic-translinear (DTL) circuit techniques. For both the CT and the DT case the user can choose the order of the filter. The output of the program in both cases is a state-space description of which the A matrix has an orthonormal ladder form [3], corresponding with a wavelet filter topology with close-to-optimal dynamic range, minimal power consumption and minimal sensitivity to component variations. As an example, we used the program to derive a 5th order discrete-time Gauss1 wavelet filter for cardiac signal analysis in pacemakers. The resulting wavelet filter has discrete-time integrators as active elements, runs at a sampling frequency of 300Hz and is implemented using switched capacitor circuit techniques. Its power consumption allows implementation in the sense amplifier of cardiac pacemakers to reliable detect the heart rate, even in presence of strong (50 or 60 Hz) interference. The design methodology can also be adopted for the design of electronic filters to implement any arbitrary transfer function or impulse response.
Dick Plettenburg, Gerwin Smit
Abstract: In order to meet the needs of a person with an arm defect, many, sometimes conflicting, requirements have to be fulfilled. These requirements can be summarized into three basic demands: cosmetics, comfort, and control [1, 2]. In the control domain natural, intuitive, subconscious control is strived for. To achieve this type of control proper feedback needs to be present [3]. This implies control according to the voluntary closing principle. In body-powered, body-controlled prostheses voluntary closing control enables proprioceptive feedback of position, velocity, and force to the user through the bodies own proprioceptive system, comparable to the way we use a tool, e.g. a tennis racket [4, 5]. The Delft Institute of Prosthetics and Orthotics has started a project to develop body-powered, body-controlled voluntary closing hand prostheses. Ultimately, we aim at voluntary closing hands that provide adaptive grasp patterns. Adaptivity enhances the natural appearance [cosmetics], and at the same time reduces the need for high pinch forces [comfort, control].
Gwen Mulder
Abstract: Cerebral or intracranial aneurysms are localized, thin walled dilatations of the arteries in the brain. The major risk involved is rupture of the aneurysm, resulting in a subarachnoid hemorrhage. In order to better decide whether or not and how to perform clinical intervention, the risk of rupture should be determined more accurately than currently possible. It is hypothesized that the risk of rupture is related to aneurysm geometry as well as to local hemodynamics. The aim of our research is to analyse intra-aneurysmal vortex patterns and identify candidate hemodynamical parameters (e.g. vortex strength) that are predictive for rupture risk and hereby improve its clinical assessment. \paragraph{Methods} To obtain patient specific hemodynamic parameters, x-ray based methods to measure the flow in the parent artery \textit{in vivo} are being developed. In order to validate these methods \textit{in vitro} experiments are carried out. Hereto, both x-ray and particle image velocimetry (PIV) experiments have been performed in idealized and patient specific aneurysm models. In addition to the experimental methods, patient specific computational methods, based on finite element approximations, are used to determine the intra-aneurysmal velocity field. Boundary conditions are based on the flow data and 3D geometries from the \textit{in vitro} experiments. This yields additional information required for the analysis of the 3D velocity fields. Hereto, suitable vortex identification schemes are implemented and used to identify candidate clinical parameters. \paragraph{Results} A computational model based on a standard Galerkin finite element approximation and an Euler implicit time integration has been applied and has been validated using the PIV results. The intra-aneurysmal velocity fields within both idealized and patient specific geometries reveal complicated 3D velocity fields (see figure \ref{fig:cfd}) with complex vortical structures. In order to analyze these vortices, a vortex identification scheme based on the second eigenvalue of $\bf{S}^2 + \boldsymbol{\Omega}^2$ is applied. Here $\bf{S}$ and $\boldsymbol{\Omega}$ represent the symmetric and anti-symmetric part of the velocity gradient tensor respectively [1]. \paragraph{Conclusions} In order to gain better understanding of the hemodynamical factors involved in intra-aneurysmal flows, a thorough analysis of the velocity field within an idealized geometry is performed using both experimental and computational methods. In the presentation the focus will be on the CFD model and the analysis of the intra-aneurysmal vortices.
Eduard van der Houwen, Bart Verkerke
Abstract: Laryngectomised patients and other patients with ineffective or missing vocal folds are very much hampered in their verbal communication. Common aids or techniques for voice generation or restoration are the use of oesophageal speech or external devices such as the ServoxTM electro larynx (Servox AG, Troisdorf, Germany). Endeavours are made into the development of improved oesophageal speech by means of surgical techniques and implanted tracheo-oesophageal shunts of which the latter can be combined with voice producing elements within [1]. Also some form of implanted mechanical or electronic voice prostheses have been applied in the past with mixed results [2,3,4]. This project aims at developing a (partly) implanted, electronic voice prosthesis based upon application of state-of-the-art, common-of-the-shelf (cots) technology, based upon the idea that medical device technology should benefit from advances in commercial mass-produced technology where ever possible. The project is divided in a 4 phase back-to-front approach, starting with the physical sound generation, working its way back to human-interfacing and energy supply (fig. 1). Current focus is on energy-efficient algorithms that generate a minimum of the required spectral components, maximizing intelligibility. Up to now a miniature amplifier generating 2.4 W ([4] reports 1.44 W to be sufficient), and creating a maximum of 60 harmonics (for vowels) and a freely configurable noise window (for consonants) is realised. Preliminary results using sub-optimal settings show that intelligibility and sound pressure levels compares to available electro-larynx systems. Main advantage over present artificial voice expedients will be the overall intelligibility (loudness and intonation) of the generated voice and the discreteness of use.
Neda Sepasian, Anna Vilanova, Luc Florack, Bart ter Haar Romeny
Abstract: We present multi-valued solution algorithm for geodesic based fiber tracking in a tensor-warped space given by diffusion tensor imaging data. This technique is based on solving ordinary differential equations describing geodesics by a ray tracing algorithm. The algorithm can capture all possible geodesics connecting two given points instead of a single geodesic captured by Hamilton-Jacobi based algorithms. Once the geodesics have been computed, using suitable connectivity measures, we can choose among all solutions the most likely connection pathways which correspond best to the underlying real fibrous structures. In comparison with other approaches, our algorithm gives the possibility of applying different cost functions in a fast post processing. Moreover, the algorithm can be used for capturing possible multi-path connections between two points that can happen when, e.g., pathologies are presented. Synthetic second order diffusion tensor data in a two dimensional space are employed to illustrate the potential applications of the algorithm to fiber tracking.
Gang Wu, Ernst Hunziker, Daniel Wismeijer, Yuelian maria Liu
Abstract: Objectives: The effective healing of bony defects − particularly of voluminous ones − remains an unsolved problem in implantology and orthopedics. To address the problem, Bio-Oss® was rendered osteoinductive and more biocompatible by coating it with a layer of calcium phosphate (CaP) with an osteoinductive agent, BMP-2, incorporated. Methods: Bio-Oss® granules (Geistlich) 0.5 -1.0mm in diameter, were coated with a layer of CaP according to our standard, two-stage biomimetic procedure. BMP-2 (10µg/ml) was introduced into the coating medium (coating inc. BMP-2). CaP-coated Bio-Oss® bearing no BMP-2 (Coating only), and uncoated Bio-Oss® bearing either no BMP-2 (Bio-Oss® alone) or an adsorbed depot of the agent (adsorbed BMP-2), served as controls. Samples were implanted subcutaneously in rats and retrieved at 5 weeks post-operation. The volume density of newly formed bone and bone marrow tissue were used to indicate the osteoinductivity and short-term biocompatibility. Volume densities of foreign body giant cell (FBGC) and fibrous capsule were evaluated as an indicator for inflammatory response and long-term biocompatibility respectively. Results: The volume densities of both bone and bone marrow in group of coating inc. BMP-2 were significantly higher than BMP-2 adsorbed groups, whereas the two tissues could not be found in groups of coating only and Bio-Oss® alone. Volume density of FBGC in the group of coating inc. BMP-2 was significantly lower than the groups of adsorbed BMP-2 and Bio-Oss® alone. Similarly, volume density of fibrous capsule in the group of coating inc. BMP-2 was also significantly lower than the group s of coating only, adsorbed BMP-2 and Bio-Oss® alone. Conclusion: BMP-2 incorporated into biomimetica CaP coating onto Bio-Oss® could not only produce bone in a large amount of good quality, but could also enhance biocompatibility and down-regulate the host inflammatory reaction.
Bogdan Necula, Lidy Fratila-Apachitei, Iulian Apachitei, Jurek Duszczyk
Abstract: The major problems encountered with total joint replacements such as implant associated infections and loosening due to poor fixation pose new challenges in developing implant surfaces with improved and multiple biofunctionalities. The synthesis of a novel composite coating consisting of a porous oxide layer bearing hydroxyapatite nanoparticles as osteoconductive component and silver nanoparticles as antibacterial agent is presented in this study. The aim was to assess the feasibility of the plasma electrolytic oxidation process to produce such a coating on a Ti-6Al-7Nb medical alloy in a single step process. The composite coatings were produced using an electrolyte bearing the two types of nanoparticles (silver and hydroxyapatite). Plasma electrolytic oxidation was performed under both galvanostatic and potentiostatic conditions. Different oxidation durations and particle concentrations in the electrolyte were chosen in order to find the optimum conditions for oxide layer growth and simultaneous particles incorporation. The surface of the composite coatings was investigated by Scanning Electron Microscopy (morphology), X-Ray Fluorescence Spectroscopy (chemical composition), Laser Scanning Confocal Microscopy (surface roughness) and by Drop Shape Analyzer (surface wettability). In parallel, the in-vitro antibacterial activity against Methicillin-resistant Staphylococcus aureus and preliminary in-vitro human fetal osteoblast cell response were assessed for the coatings bearing a single type of particle i.e., silver and hydroxyapatite, respectively. Depending on the process parameters used, different porous oxide layers with pore sizes ranging from few hundred of nanometers up to 10 µm, bearing both types of nanoparticles, hydroxyapatite and silver, were successfully formed on the titanium medical alloy. The composite coatings revealed an increased roughness compared to the non-oxidized metal and an improved wettability relative to the specimens oxidized in the absence of nanoparticles in the electrolyte. Based on the results obtained, plasma electrolytic oxidation can be considered a technically feasible method for multiple biofunctionalization of titanium medical alloys. In-vitro antibacterial testing of silver bearing coatings showed a highly effective bactericidal activity confirmed by a complete killing of S. aureus within 24 hours as oppose to the untreated surfaces. The preliminary results of the human osteoblast cells cultured on the hydroxyapatite bearing coatings indicated colonization of the disks by the cells and production of extracellular matrix. On-going work is focused on the cytocompatibility and antibacterial properties of the coatings bearing the two types of particles and on the control of nanoparticles incorporation during the oxidation process.
Heidi Witteveen, Marit van Weering, Laura Kallenberg, Miriam Vollenbroek-Hutten, Hermie Hermens
Abstract: Chronic low back pain (CLBP) is a common problem in the Western industrialized countries. Not only because of the negative influence on a patient’s life, but also because of the high medical and social costs. Treatment of CLBP in rehabilitation centers comprises mental, social, and physical aspects. Several studies have shown that muscle activation patterns of CLBP patients differ from those of healthy subjects [1]. However, these studies have been performed in a laboratory environment and during short term measurements not reflecting the patients muscle behavior in every day life. The aim of our study was to investigate the muscle activation patterns of CLBP patients in their daily life and compare these to asymptomatic controls. This information will provide a basis for the development of a feedback application for CBLP patients. For the measurement of electromyographic (EMG) activity of the lower back muscles in a comfortable way and with a high quality of signals, a personally adjustable garment with integrated electrode holders was developed. Surface EMG (of the m. Longissimus Dorsi) was measured of 13 CLBP patients and 14 healthy controls during two days, in the morning (4 hours) and in the evening (4 hours). Muscle activation patterns were analyzed using RMS values (level of muscle activity), the coefficient of variation in muscle activity (mean/standard deviation), and the RRT values (the percentage of RMS blocks < 5 µV over 15 minutes) Furthermore the course of RMS over a day was plotted. Before the start of the measurements, the Tampa score (level of kineasiophobia) of the patients was assessed to characterize the patient group. Analysis of the calculated parameters showed no significant differences between the group of CLBP patients and healthy controls. However, a more detailed analysis showed some first indications that the patient group can be divided into a group showing low RMS values and a high RRT, and another group with a low RRT and low values of variation compared to the controls. Even with the small numbers of this study the difference in mean RMS between the first patient group and the control group (13.99±4.61 vs. 20.42±6.03 µV) and the difference in mean RRT between the second patient group and the controls (20.83±5.73 vs. 33.73±12.71%) were (almost) significant (p=0.07 and p=0.02). The patient groups differ in mean Tampa score (34.4±3.29 vs. 26.5±5.18, p=0.02); High Tampa scores associated with low RMS values, while low Tampa scores associated with low RRT values. The results show that a part of the patients show lower muscle activation patterns while the other part shows less variable and higher muscle activation patterns compared to controls. Considering myofeedback to ‘normalize’ muscle activation patterns of CLBP patients, our results indicate the use of different strategies for both patient groups: The first group (low RMS) should be stimulated to undertake more activities, while the second group (low RRT) needs to learn to alternate between activation and relaxation by increasing the amount of rest periods.
Martijn van de Giessen, Geert Streekstra, Simon Strackee, Mario Maas, Kees Grimbergen, Lucas van Vliet, Frans Vos
Abstract: Comparing wrist shapes of different individuals requires alignment of these wrists into the same pose. Unconstrained registration of the carpal bones results in anatomically non-feasible wrists. In this paper we propose to constrain the registration using the shapes of adjacent bones, by keeping the width of the gap between adjacent bones constant. The registration is formulated as an optimization involving two terms. One term aligns the wrist bones by minimizing the distances between corresponding bone surfaces. The second term constrains the registration by minimizing the distances between adjacent sliding surfaces. The registration is based on the Iterative Closest Point algorithm. All bones are registered concurrently so that no bias is introduced towards any of the bones. The proposed registration method delivers anatomically correct configurations of the bones. The registration errors are in the order of the voxel size of the acquired CT data (0.3x0.3x0.3mm3). The standard deviation in the widths of gaps between adjacent bones is in the order of 10% with an insignificant bias. This is a large improvement over the standard deviations of 30% to 80% encountered in unconstrained registration. The value of this method is its capability of accurately registering joints in varying poses resulting in physiological joint configurations. In future research we will concentrate on building a statistical shape model that preserves physiological spatial relations between adjacent wrist bones. The training set of wrists in this model will be registered to a single pose with the described constrained registration method.
Han van Triest, Marc van Zandvoort, Hans van Assen, Bart ter Haar Romeny
Abstract: n the western world, vascular disorders form a major medical problem. To increase knowledge of the underlying mechanisms of, for example, atherosclerosis, extensive research is performed. At the university of Maastricht, a special type of microscopy is used for this task, called Two Photon Laser Scanning Microscopy (TPLSM). Using TPLSM, three dimensional images can be extracted from viable arteries [1]. To describe the processes occurring in the large arteries as a reaction on changing circumstances, a method is proposed for the estimation of the radius of vessels from TPLSM-images. The proposed algorithm is based on the Hough Transform; an evidence gathering algorithm that operates by means of voting for the best set of parameters. It is assumed that the cross-section of the vessel visible in the TPLSM-data represents a circle. Each combination of three non-co-linear points uniquely define a circle. In the proposed algorithm, each combination of three points vote for a potential center (xc, yc). Since a combination of points that lie far from each other give a more accurate fit, the vote is weighed by the distances between P1, P2 and P3. Experiments: To validate the proposed method, ten images are used. Images contain both parts of the wall, i.e. upper and lower part. This is needed for the determination of the true diameter of the vessel. Ground truth is formed by the manual estimation by 12 volunteers. The results are compared with a standard Least Squares Estimator (LSE) due to Zhou et al. [2]. Results: We display the results of the proposed algorithm and the LSE against the number of points used for the fits. We show the average errors of the fits expressed in percents, and the standard deviation of the error. Votes are stored in a two dimensional array, in which the best fit is represented by the coordinates of the peak present in parameter space. Conclusions: A method has been proposed for the estimation of the vascular diameter that, when using at least 100 data points, results in more stable estimates, compared to the standard Least Squares method. The standard deviation of the error is reduced by a factor of up to 5 times.
Willemijn Groenendaal, Natal van Riel, Jeroen Jeneson, Klaas Nicolay, Peter Hilbers, Robert Wiseman
Abstract: Sites of calcium release, uptake and action are highly organized and densely packed in skeletal muscle cells. This organization suggests an important role for the spatial distribution of organelles and calcium sensitive proteins in muscle function. A computational model was developed [1], to be able to study the influence of this organization on skeletal muscle behavior. For model parameterization and validation, data obtained at physiological temperature is necessary. However, measurements in skeletal muscles are usually performed at unphysiological temperatures (4-25°C). To be able to validate the model and consequently to translate this data to physiological temperature, the influence of temperature on skeletal muscle behavior should be quantified. Calcium fluorescent dye experiments and force measurements were performed to quantify the influence of temperature on skeletal muscle dynamics. Murine extensor digitorum longus (EDL) muscles were isolated and mounted in the set-up. Muscle length was adjusted to optimal resting length and stimulation voltage was adjusted to generate maximal force. Rhod-2 AM was loaded in whole muscle at 37ºC for 30 minutes or until force was decreased with >25%. After the loading period the muscle bath was washed at least 5 times. Next muscles were stimulated at different temperatures (range 15-20-25-30-35ºC) and different frequencies (temperature dependent, twitch to fused tetanus). Fluorescence and force traces were recorded simultaneously with Felix (DeltaRAM; Photon Technology International). Separately, force kinetics, also at physiological stimulation frequency, in the absence of fluorescent dye were measured as in ter Veld et al [2]. In this way we were able to characterize the influence of fluorescent dye on the force kinetics. All data were analyzed in Matlab 7.1 (the Mathworks Inc). The following characteristics were determined: rise and decay time, half width, area under the curve, and peak value. The relation between temperature and characteristics was quantified, e.g. the rise and decay time of the fluorescence and force signals decreased, when the temperature increased. The relations were found to be significant (one-way ANOVA, p-value<0.05). Q10 (a measure of the rate of change of a biological or chemical system as a consequence of increasing the temperature) values were between 1.5 and 3.2, within the expected physiological range. A statistically significant non-linear relation between temperature and calcium and force dynamics was determined. Thus to be able to make physiological relevant conclusions, data obtained at low temperatures should first be translated according to these relations to physiological temperatures. In the future, the model can assist in this conversion. References: 1: Groenendaal et al. IET Systems Biology, in press, 2: Ter Veld et al Pflugers Archive 2006
Tim Peeters, Vesna Prckovska, Markus van Almsick, Bart ter Haar Romeny
Abstract: High Angular Resolution Diffusion Imaging (HARDI) is an emerging Magnetic Resonance Imaging (MRI) technique that overcomes some decisive limitation of its predecessor Diffusion Tensor Imaging (DTI). HARDI can resolve locally more than one direction in the diffusive pattern of water molecules and thereby opens up the opportunity to display and track crossing fibers. Showing the local structure of the reconstructed probability profiles in a fast, detailed, and interactive way can improve the quality of the research in this area and help to move it into clinical application. In this paper we present a novel approach for HARDI glyph visualization or, more generally, for the visualization of any function that resides on a sphere and can be represented by Laplace series. Our GPU-accelerated glyph rendering improves the performance of the traditional way of HARDI glyph visualization as well as the visual quality of the reconstructed data, thus offering interactive HARDI data exploration of the local structure of the white brain matter in-vivo. In this paper we exploit the capabilities of modern GPUs to overcome the large and memory-consuming data visualization.
Carole Leguy, Marielle Bosboom, Hanneke Gelderblom, Arnold Hoeks, Frans van de Vosse
Abstract: Background: Arterial stiffness, S, is an independent predictor of cardiovascular risk at an early stage, S being defined as the product of the wall thickness, h, and the Young modulus, E. Objective: The goal of this study is to investigate the feasibility of a new method based a one-dimensional wave propagation model to estimates S. Method: A reverse method based on a patient specific wave propagation model of the upper limb was used. To estimate the initial model parameters, local ultrasound measurements as well as the systolic and diastolic blood pressure (BP) were obtained from a group of 6 healthy volunteers. The vessel wall distension and blood velocity were assessed at 5 positions along the arm, the blood volume flow (BVF) was then estimated from the centreline velocity using the Womersley profile. The brachial, ulnar and radial arteries were modeled as linear elastic with increasing S and exponential decay of the diameter toward the distal part. At the more distal positions, the end-impedance Winkessel parameters Z0, Rv and Cv were obtained from a fitting of the BVF and WD waveform at the radial and ulnar arteries. At the more proximal site, the BVF, , was used as an input for the simulations. Along the arterial tree, a distributed outflow, , was estimated from the measured time average BVF. Then, the simulated BVF and BP curves were compared with the in-vivo results. A reverse method was used to adapt the model parameters resulting in values reflecting physiological results. Results: Simulated BVF and BP waveform resulting from the iterative method are fitting well the in-vivo estimates at the brachial, radial and ulnar arteries. The obtained S in the brachial artery equals 0.34±0.08 kPa.m and is 40% lower than the in-vivo estimated S from the brachial artery distensibility 0.57 ± 1.3 kPa.m. Conclusion: Patient specific wave propagation models can be used to improve the estimation of in-vivo arterial stiffness. However, differences are observed between simulated BP and in-vivo assessed wall distension curves. We hypothesize that those differences are due to the visco-elastic properties, therefore those properties are currently being implemented in the wave propagation model.
Christian Tiemann, Natal van Riel, Kasper Rouschop, Maud Starmans, Philippe Lambin, Peter Hilbers
Abstract: Hypoxia is a reduction in the normal level of tissue oxygen tension and is a common feature in many diseases including cancer [1]. Hypoxia triggers several oxygen-sensing pathways that alter cell metabolism and allow the tumour to adapt to the hypoxic micro-environment. Hypoxic cells display a high resistance against conventional therapies, like radiotherapy and chemotherapy. Therefore, these hypoxia induced pathways have become of increasing interest for additional tumour therapies. The best studied hypoxia-response is driven by a transcriptional program initiated by (regulated) stabilization of hypoxia inducible factor 1 (HIF-1). The HIF-1 response is complex and depends highly on both the severity and duration of hypoxia. Moreover, the response shows cell-type specific characteristics. Understanding the dynamic behaviour of the HIF-1 response will facilitate future progress in targeting tumour-hypoxia. However, understanding a complex bio-regulatory network is often difficult and therefore computational models are becoming of increasing interest. Past years, several computational models of the HIF-1 pathway have been presented in literature. To our knowledge, Kohn et al. [2] were the first to publish a computational model. Next, Yu et al. [3] simplified their model. They both used a qualitative approach to model the HIF-1 response and successfully described a ‘switch-like’ oxygen-response of HIF-1 activity. However, experimental data of Jiang et al. [4] showed an ‘exponential-like’ oxygen-response of HIF-1 activity. The aim of this study is to develop a computational model using a quantitative approach which is able to reproduce current experimental data. A minimal model of the HIF-1 response has been developed, containing eight molecular species and nine unknown model parameters. The parameters have been estimated by minimizing the difference between data of Jewell at al. [5] and the model output using a least squares criterion. Furthermore, the identifiability of the network components has been studied extensively by means of Fisher Information Matrix (FIM) analysis and a Monte Carlo approach. A Local Parametric Sensitivity Analysis (LPSA) and Multi Parametric Sensitivity Analysis (MPSA) have been carried out to assess model robustness and revealed key-sensitive model components. This has contributed to the optimal design of new experiments to improve model reliability. To validate the model, we demonstrate that the model is capable of reproducing two other data-sets of Jewell et al. [5] and Jiang et al. [4]. This has lead to new insight in which parts of the pathway are specifically oxygen-sensitive and indicates that the complex formation of HIF-1α and PHD is the oxygen-sensitive step rather than the actual hydroxylation of HIF-1α. This hypothesis is open for experimental validation.
Sander Schutte, Frans Goudsmit, Fred van Keulen, Frans van der Helm, Huib Simonsz
Abstract: Purpose: To reach a full understanding of the suspension of the eye, we developed a model of orbital mechanics based upon FE modeling and in vivo data. In most studies the way the eye is suspended in the orbit remains undefined, i.e. the center of rotation of the eye is fixed and distributed loads are reduced to single force vectors. The focus of this study was on (1) implementation of the oblique muscles and (2) the optic nerve and on (3) validation using MRI data. Methods: Contours of bony orbit, eye muscles and eye were determined from MRI. Mathematical surfaces were fitted through the contours to describe the 3D-boundaries of the six extraocular muscles, the optic nerve and the eye. A tetrahedral mesh was constructed in each structure and in the remaining volume, the latter comprising fat including connective-tissue septa. Spatial inhomogenities were extended and at several points sliding surfaces were introduced, e.g. tendons over sclera, medial rectus over orbital wall and Tenon's capsule over sclera. An Optical Flow algorithm was used to estimate deformation of the orbital fat. Results: Simulations of active horizontal eye movement, of single-point passive forced duction and dual-point passive forced duction yielded different centers of rotation and different values of stiffness, largely corresponding to literature data. Torsional forced duction showed no sideways displacement of the muscle belly at the level of the posterior pole, despite the lack of explicit connective-tissue structures to keep the muscle bellies in place. The predicted deformation of the fat roughly corresponded to that found with MRI. Conclusions: FE modeling offers unique possibilities to examine complex disorders of eye motility and enables a full understanding of the suspension of the eye in the orbit. Three dimensional material properties of the muscles have to be measured, for a sound basis of the model.
Dirk-Jan Kroon, Kees Slump
Abstract: Medical image data often consist of soft uniform regions with elastic properties and with large inter-patient variations. So it make sense to use a non-rigid registration method for medical data which is not based on landmarks, and to use a registration method which smoothly deforms the data. This characterizes our registration method which is based on a grid of B-splines which smoothly deforms an image to maximize similarity between two images. The basis of our method is a free form deformation (FFD) grid of 1D B-splines which was introduced by Rueckert et al. [1]. Control points of the b-spline grid are moved to transform the image locally and smooth. A similarity measure between target image and the control point transformed image is used to measure registration improvement. The optimizer needs a similarity gradient to move the control points, we determine the gradient by central differences. The transformation of the control points is based on local registration clues, to improve the global registration a penalty based on the bending energy of a thin sheet of metal is added. The optimizer will iterative minimize the registration error and maximize the similarity between the images. Rueckert uses a steepest gradient optimizer in combination with mutual information as similarity measure. We compare different similarity measures for speed and accuracy, the sum of squared difference (SSD), normalized mutual information (NMI) and Gradient Difference (GD). Also we compare different optimizers, the steepest gradient optimizer, Quasi-Newton method and Levenberg-Marquardt nonlinear minimization. To speed up, our method uses grid and image refinement, polynomial look up tables, and uses the fact that one control point only influences a part of the image. We did a comparison test between B-spline registration and demon registration [2], which is a type of fluid registration. We tested the performance on two 256x256 deformed Lena images without noise. The B-spline registration is very accurate with a SSD error of 21.7 after registration, and registration time of 346 seconds on a desktop computer. Demon registration is also accurate with a SSD of 18.1, and needed only 25 seconds to register the image. B-spline registration also gave accurate registration results when registering T1 atlases and white matter probability atlases to MRI patient data, with the use of mutual information. Future work will focus on registration accuracy with noise and bias fields, and including landmark information from SIFT as a penalty in this non-rigid registration method.
Heleen van Beusekom, Wim van der Giessen
Abstract: Coronary Imaging e.g. IVUS and OCT to assess vascular size, plaque morphology and composition are useful tools in clinical trials as well as routine practice. It is unclear whether imaging catheters induce acute endothelial cell damage (ECD) and how this heals over time. Methods. Normal porcine coronary arteries were subjected to intravascular imaging (Raman, IVUS, OCT), only guidewire placement or no manipulation (control). ECD was assessed acute or at 5 days (chronic) with an Evans Blue dye-exclusion test and planimetry followed by scanning EM (SEM). Results. Acute injury (% staining of the area exposed to imaging) was considerable, but similar for the 3 imaging catheters. Guidewire-only induced less ECD. One-way ANOVA showed that only OCT- and RAMAN induced more ECD than control. SEM showed that staining was associated with either EC denudation or injured but present EC. At 5 days staining was reduced to background levels. Conclusion. Vascular imaging and guide wire manipulation both induce acute ECD, but heals within 5 days following intervention. A,Raman;B,IVUS;C,guidewire;D,control Planimetry of endothelial injury injury % acute (n) % chronic (n) Raman 38±21* (11) 2.1±2 (3) OCT 43±15** (4) n.d. IVUS 30±12 (3) 1.5±0.9 (2) Guide wire 21±13 (4) n.d. control 2±2 (3) 0.7±0.6 (3) * p=0.002 and **0.02 vs. control
René van Oers, Bert van Rietbergen, Peter Hilbers, Keita Ito, Rik Huiskes
Abstract: The remodeling of trabecular bone is carried out by osteoclasts, digging trenches across the trabecular surface, and osteoblasts, filling these trenches with new bone. It is unknown why osteoclasts resorb along the surface and do not perforate the trabecula. Smit and Burger [1] performed a finite element analysis of a trabecula with a resorption lacuna. They found that strains were highest at the bottom of the lacuna. We hypothesize that these strains induce osteocytes (cells within the bone) to secrete osteoclast-inhibiting signals, thus restricting osteoclastic resorption to the trabecular surface and preventing perforation. Earlier, we developed a finite-element based bone adaptation model [2], that produces trabecular-like architectures. Depending on the local strain-energy-density osteocytes send signals to the bone surface, where they control the activity of osteoclasts and osteoblasts. Osteoclastic resorption was represented as cavities occurring along the bone surface. For the present work osteoclasts are explicitly represented with a cell simulation model [3], which allows us to investigate how the proposed strain-based signals guide resorbing osteoclasts. An osteoclast is placed at the side of a single trabecula which is loaded in its longitudinal direction (Fig.1 left). In the loaded situation, the osteoclast meanders along the surface (Fig.1 middle) since its resorption depth is limited by strain-induced inhibiting signals that are strongest at the bottom of the trench. When the trabecula is not loaded, resorption is random (Fig.1 right), and likely to perforate the trabecula. These results demonstrate that the typical surface resorption of osteoclasts can indeed be explained by mechanical factors.
Marion Geerligs, Lambert Breemen, Paul Ackermans, Gerrit Peters, Cees Oomens, Frank Baaijens
Abstract: Introduction: Knowledge about the mechanical behavior of the upper part of the skin is required to improve skin-device contact. Mechanical properties of entire skin and skin layers have therefore been studied using a broad variety of testing techniques. Until today, however, no consistency exists on the mechanical properties of the upper skin layers: e.g. the epidermis and its top layer, the stratum corneum [Fig. A and B]. Due to the complex structure of skin, it appeared that the determination of the contribution of each skin layer in vivo is difficult. This study therefore aims to determine the contribution of epidermis and stratum corneum on the overall skin response using an in vitro indentation method. Methods: Human abdominal skin was obtained from the plastic surgery department. Samples of individual skin layers as well as skin samples with various thicknesses were prepared using mechanical and/or chemical methods. Indentation experiments were performed on a MTS NanoIndenter XP system. Measurement locations were manually selected under the microscope [Fig C]. The thickness of the samples determined the diameter of the spherical indenter tip, either 0.5 or 2 mm, and the maximum applied load. The loading ramp always lasted 100 s and the maximum load was maintained 30 s. From the initial unloading curve, the reduced modulus ER was calculated. Results: Reproducible results were found for stratum corneum, epidermis (100 μm) and epidermis with dermis (thickness 200 and 500 μm). For all individual skin layers as well as the different skin samples, the reduced modulus could be calculated. The stiffness for the epidermis was 0.4 MPa. Discussions: This study introduced a new indentation method for skin. The chosen method is known for its accuracy and reliability. Therefore, this technique is an interesting method towards a reliable characterization of skin. The measurement system has been adapted for soft biological materials. To our knowledge, this has been the first time that the stiffness of epidermis exclusively has been determined.
Frits Hermans, Rob Heethaar
Abstract: Knowledge about the current distribution in the human body is crucial for diagnostic techniques like electrical impedance tomography and cardiography. The current that flows through the lungs and the heart are the variables that are used to calculate parameters about either lung or heart function. The aim of this research was to determine which portion of the total electrical current flows through the heart and the lungs. The effect of geometry changes due to breathing and the heart contraction on the electrical current density distribution through the body was investigated. The influence of changes in electrical resistivities of the lungs in inspiratory and expiratory states on the current density distribution was determined. We used 2D and 3D finite elements models (FEM) of the human thorax to determine the current density distribution through the lungs, the heart and the aorta in four-electrodes configurations. The Visible Human Project (VHP) data set was used for the coronal 2D simulation. MRI images created from a healthy subject in inspiratory and expiratory states during one heart cycle formed the basis for the 3D geometry, by segmentation with the SimpleWare computer program. For the FEM analysis the Conductive Media DC module of Comsol Multiphysics was used. Two electrodes placed on the laterolateral sides of the neck in the FEM model delivered an inward current flow and two electrodes on the laterolateral sides of the thorax, at the heigth of the 6th rib conduce an outward current flow. Electrical resistivities were assigned to different tissues as found in literature [1]. In the 2D model, breathing was simulated by changing the electrical resistivity from the inspiratory state to the expiratory state. The diastolic state differed from the systolic state by changing the width of the aorta by 15%. On a horizontal line through the heart and the lungs the current density was determined by FEM analysis in the four states as described. In the 3D model, the geometry of the lungs in inspiratory and expiratory state could accurately be determined by segmenting 2D slices and creating 3D geometries. As in the 2D simulation, the electrical resistivities of the lungs in inspiratory and expiratory states were distinguished. The portion of the electrical current density through the lungs and the heart was determined by a plane through the lungs and the heart. The models set-up allowed us to calculate current distributions through organs of interest in the human thorax through arbitrarily chosen planes or cross sections based upon anatomical data of the subject under study.
Herman Schepers, Peter Veltink
Abstract: Traditionally, human body motion tracking is done in gait laboratories equipped with several measurements systems. Body movement is measured by an optical position measurement system, ground reaction forces by a force plate mounted in the floor, and muscle activity using EMG. The major drawback of these systems is the restriction to a laboratory environment, which inspired several research groups to start initiatives for the development of ambulatory alternatives. A common used ambulatory alternative for the laboratory fixed optical tracking system is to use inertial and magnetic sensors. An inertial and magnetic sensor consists of a 3D accelerometer, a 3D gyroscope and a 3D magnetometer. Position and orientation can be estimated by integration of the signals measured by the sensors. However, the integration introduces drift caused by noise and a fluctuating offset. Moreover, there is no information available about the relative position and orientation of sensors with respect to each other. To overcome these limitations, inertial sensing can be fused with an aiding system. In this study, inertial measurements are fused with magnetic measurements. The magnetic field is generated by an on-body magnetic actuator and measured by a 3D magnetometer [1]. Although an example of the fusion of inertial and magnetic tracking has been described in a previous study [2], several aspects need to be improved. The described fusion algorithm, based on a magnetic dipole approximation of the source, requires all three source coils to be actuated every update, while the update rate is fixed and pre-assigned by the user. Moreover, the source consisted of three coils which need to be mounted orthogonally and share the same origin. In order to apply the ambulatory magnetic tracking system more generally, an estimation method is proposed which is not based on the dipole approximation and is based on a single coil together with a 3D inertial and magnetic sensor. The method allows the measurement system to achieve maximal accuracy at minimal energy consumption. The purpose is to actuate only when the uncertainty associated with the estimated position and orientation exceeds a pre-defined threshold. Moreover, only the coil delivering most information is actuated, while the current applied is such that the uncertainty associated with the position and orientation decreases as desired. The objective of this study is to propose and evaluate a part of the estimation method, a stochastic measurement model. The measurement model is used to predict the magnetic field measured by the sensor, based on an analytical expression for the magnetic field generated by the source coil evaluated at the current estimation of position and orientation. The prediction will be compared to the actual field measured by the sensor, and the difference is used to correct the estimation of position and orientation. In a successive study, the measurement model will be used in an adaptive filter structure, allowing maximal accuracy at minimal energy consumption.
Alina van der Giessen, Harald Groen, Jolanda Wentzel, Antonius van der Steen, Pim de Feyter, Frans van de Vosse, Frank Gijsen
Abstract: Nowadays, it is possible with multi-slice computed tomography to assess non-invasively 3D patient-specific geometries of the coronary tree. These 3D geometries include the side-branches of the main vessels, thus wall shear stress (WSS) calculations near bifurcations are now feasible in order to study the relationship between WSS and atherosclerosis. The outflow conditions (OC) for the side-branches determine the flow rate through them and thus greatly influence the WSS computations in the main vessel. In this study we investigate the difference in WSS distribution when prescribing three different OC models. Ten patients with coronary artery disease were scanned with a 64-slice MSCT (Siemens, Germany) scanner. Each main branch of the coronary tree with a part of its side-branches was segmented with the Vascular Modeling Toolkit [1] and meshed for the calculations. We simulated a 3D, incompressible and steady flow in a rigid geometry. The blood was assumed to be non-Newtonian (Carreau-model). The inflow was chosen such that the WSS at the inlet was 0.68 Pa. We solved the Navier-Stokes equations and assessed the WSS with the three different OC on the outflow tracts for each geometry: 1) zero-stress, 2) flow fraction, expected based on Murray’s Law, 3) flow fraction expected based on a stochastic 1D model of the coronary tree generated as by Dankelman et al [2]. For zero-stress outlet, flow through the side-branches is determined by their resistance. Since only a small part of the side-branches is modeled, the first large side-branches has a very low resistance, resulting in unrealistic low flows and WSS in the main branch. Murray’s Law and the stochastic 1 D model predict different flow rates though the small side-branches: the stochastic model generally predicts higher flow rates through small side-branches (see figure). If we consider segments between side-branches the relative differences in WSS between the Murray’s Law model and the stochastic model are relatively constant and they are equal to the relative difference in flow through the main branch in between the side-branches. Near the side-branches, the location where atherosclerosis occurs most frequently, differences between the two models are larger; in the example shown up to 40%. The two more realistic outflow models based on Murray’s Law and the 1D stochastic model result in different WSS values in the main branch. In segments between side-branches the different is WSS values for the two models are relative to the differences in flow. Near bifurcations, significant non-uniform effects of the models on the WSS distribution were established. The different models should therefore be validated against flow measurement before patient-specific WSS patterns can be related to atherosclerosis near side-branches.
Dyah Dewi, Bart Verkerke
Abstract: Scoliosis is a three-dimensional deformity of the spine characterized by lateral deviation and axial rotation of the vertebrae. It is important to evaluate the scoliosis progression frequently to be in time for treatment as it may become severe when the degree of curvature is increasing extremely. Spinal X-ray radiographs in posterior anterior (PA) and lateral (LA) projection are employed to determine the curvature of the spine. An accurate and reproducible measurement play a major role in the success of scoliosis management. Progression of deformity of the vertebral column is determined by observing X-ray radiographs at different time. Regarding the production of the radiographs, unstandardized radiographs in term of posture and position could create measurement errors [1]. Therefore a device to gain a reproducible posture is required in following the scoliosis progression. A good reference for the reproducible posture can be found using center of gravity of the body where the body obtains its balance [2]. Der Wippe, a balance positioning device, was developed to achieve a posture in which the subject is always forced to an identical position [3]. This device consists of flat squared block on the top and smaller cube attached in the center point at the bottom side. The subject reaches the balance by taking a natural stance with his feet on Der Wippe. In this position, the center of gravity is assumed to be constant. The balancing effect causes the subject’s muscles moving to the back to stretch and the upper body forced in an upright position. Body marks is drawn to test Der Wippe in identifying the posture. Based on translocation and movement patterns of the upper body, 8 points are drawn at both sides of T3, T7, T12, L4 of the spine in PA view and 4 points at T3, T7, T12, and L4 in LA view. Angle parameter is used to measure the movement of the points. Horizontal lines are drawn on each PA point in the left side connected to its adjacent PA point in the right side. Vertical lines are drawn to connect PA points in each left and right sides making rectangular shapes with horizontal lines. Then, 12 PA angles are created from the figure between such lines. In LA, vertical lines are drawn to connect the LA points and a horizontal line is drawn for every LA point to figure out the 6 LA angles. The angles are measured by taking picture of the subject in PA and LA sides. Twenty healthy subjects volunteered in this study. The test was done in two consecutive days in five repetitions for each day to investigate how much Der Wippe could produce reproducible posture in different time of measurement matched up to the usual way of measurement. A comparison was made between body marks angle measurement where the subject standing naturally without Der Wippe and the subject standing on Der Wippe. The experiment indicates that standing on Der Wippe provides better reproducibility than standing naturally without any supportive devices.
Marijn van Dongen, Wouter Serdijn
Abstract: Brain stimulation is a clinical treatment method that can be used for a wide variety of applications. Most current implementations of implantable brain stimulators use a generator implanted in the chest. Subcutaneous lead wires connect the generator with electrodes implanted in the brain. The fairly long wires are a common source of malfunctioning and increase the burden for the body. Therefore a stimulator small enough to be fitted inside the brain completely is desired. The biomedical electronics group of the Delft University of Technology targets the design of an implantable brain stimulator with a focus on low power consumption and small chip area. The physical principles underlying the (de)activation mechanisms of neural cells using electrical stimulation involve the elevation of the tissue potential up to a certain level. This is accomplished by injecting a certain amount of charge within a certain amount of time into the tissue. An important constraint needs to be obeyed: in order to prevent tissue damage the net injected charge needs to be zero. Many existing stimulator systems use current sources to stimulate the tissue, because in this way the injected charge is more easily controlled. The drawback is the relatively inefficient implementation of accurate current sources [1]. The structure of the new stimulator is therefore based on a voltage source, because of its efficient implementation, both in terms of power and area. In the voltage source based implementation, the current injected by the voltage source is sensed and integrated to measure and subsequently control the injected charge [2]. The absolute value of the injected charge is of no concern, due to the time- and patient variant nature of the electrode-tissue interface. Only the relative accuracy of the integrator needs to be assured. Using this setup any desired waveform for stimulation can be used, e.g. asymmetric or burst waveforms, while charge cancellation is always guaranteed. Due to the high flexibility and the efficient basic concept of voltage based stimulation, the concept can be used in any other kind of stimulation device, such as cochlear implants, pacemakers and muscle stimulators. REFERENCES [1] J. Simpson, M. Ghovanloo, “An Experimental Study of Voltage, Current and Charge Controlled Stimulation Front-End Circuitry”, IEEE International Symposium on Circuits and Systems (ISCAS), pages: 325-328, May 2007 [2] X. Fang, J. Wills, J. Granacki, J. LaCoss, A. Arakelian, J. Weiland, “Novel Charge-Metering Stimulus Amplifier for Biomimetic Implantable Prosthesis”, IEEE International Symposium on Circuits and Systems (ISCAS), pages 569–572, May 2007.
Jamal Esskili, Sjoerd Niehof
Abstract: ABSTRACT Introduction; Regional blocks are used to anesthetize a body part that will be subjected to surgery and is a widely used replacement for general anaesthetics. Traditionally methods to determine loss of sensation in the skin such as pinprick and cold stimulation are used to determine whether a regional block was applied successful. Both methods take approximately 20 to 30 minutes and do not provide accurate and objective results[1]. A successful block will anesthetize the sympathetic nerve fibres and lead to vasodilatation [2,3]. The loss of sympathetic nerve tone corresponds with loss of pain sensation. Therefore assessment methods of local regional blocks are based on measuring the change in blood flow. For example videothermography and Peripheral flow index. However, in general, videothermography is not available and PFI is only able to give an local indication of the dermatome area that is being measured. In conclusion, there is need for an easy to use early predictor of an (un-)successful local regional block. An accessible technique that addresses both the availability and easy of use is the Pulse Transit Time (PTT). PTT or rPTT (r-wave gated Pulse Transit Time) is a non-invasive method using photoplethysmography (PPG) and a standard electrocardiography (ECG). Per definition the PPT is the time that elapses between the onset of the R-wave in the electrocardiogram and the arrival of the resulting pulse pressure wave of blood, usually in the index finger, measured by a photoplethysmosensor. The time point commonly use to indicate arrival of the pressure wave in the index finger are arrival of the base of the wave, 50% increase, 75% increase or top of the wave. We have studied if PTT is an early indicator of the success or failure of an local regional block. Furthermore, we have studied the effect of choosing different point of arrival of the pressure wave on the PTT value. Method; The patients were elected to undergo surgery and receive an local regional block. We measured the PTT using a ECG and Plethysmosensor before, during and after the local anesthetics where induced. One plethysmosensor was attached to the arm that was receiving local anaesthetics and the other on the contra-lateral arm as an control measure. During the PTT measurement time synchronised video recordings where made using Observer XT (Noldus) so specific events during the intervention could be analyzed. Results; In the first 10 patients a clear increase in PTT value was observed minutes after administering a local regional block. Common artefacts in PTT where observed and have been addressed. Suggestion on definition of the arrival of the pressure wave have been made. Conclusion; It is likely PTT can be used as an early objective accurate indicator of a local regional block. Common artefact on PTT measurement are easily addressed and results in an clear increase on the accuracy of the PTT measurement. REFERENCES 1. Galvin EM, Niehof S, Medina HJ, et al. Thermographic temperature measurement compared to pinprick and cold sensation in predicting the effectiveness of regional blockades. Anesth Analg 2006; 102:598-604. 2. Valley MA, Bourke DL, Hamill MO, Raja SN. Time course of sympathetic blockade during epidural anesthesia: laser Doppler flowmetry studies of regional skin perfusion. Anesth Analg 1993; 76:289–294. 3. Sorensen J, Bengtsson M, Malmqvist EL, et al. Laser Doppler perfusion imager (LDPI)-for the assessment of skin blood flow changes following sympathetic blocks. Acta Anaesthesiol Scand 1996; 40:1145–1148.
Rachel Senden, Bernd Grimm, L. Verlaan, Kenneth Meijer, Hans Savelberg, Ide Heyligers
Abstract: An acceleration based gait test has been clinically validated for patients with knee pathologies1,2. This test was able to distinguish pathological from healthy gait, but was less able to distinguish finer functional differences. Because stair climbing is a more demanding task, it may produce more sensitive parameters. This study investigates whether an acceleration based stair test is able to distinguish demographic differences in healthy subjects and to diagnose meniscal tear (MT). The ascending and descending of five stairs at preferred speed was measured using a triaxial accelerometer (62x41x18mm; m= 53g; f= 100Hz) attached to the sacrum. Motion parameters [avg. step time up and down, step time up – step time down, irregularity up and down (difference between max. and min. step times) and asymmetry (difference between the step times of the dominant and non-dominant leg in healthy subjects or affected and non-affected leg in patients)] were derived from trunk acceleration using peak detection algorithms. Two groups were measured: A) 46 healthy subjects (19m/27f, mean age: 39yrs; range: 21-74) recording gender, height, weight, BMI and leg length were split into two subgroups (Young: 15m/16f; avg. age 25yrs; Old: 4m/11f; avg. age 67yrs). B) 37 patients (18-72yrs) who received arthroscopy suspecting MT resulting from trauma, degeneration or both following standard physical examination (e.g. McMurray). At arthroscopy the presence or absence of MT and osteoarthritis level was identified. Demographic effects on stair climbing were analysed in the healthy group. Patients were compared to this group to calculate stair test sensitivity and specificity for differentiating pathologic from healthy subjects and for diagnosing MT using threshold values. Avg. step times were higher ascending (606ms) than descending (575ms, p<0.05), a difference seen in most individuals (39/46) which is expected by the difference in energy expenditure. The step time difference between ascending and descending was 31ms with a sign. difference between the young (47ms) and elderly group taking more time to descend (-7ms). All healthy subjects with descending times 20ms slower than ascending (6/46) were found in the elderly group (p<0.05). Irregularity was nearly equal ascending (142ms) and descending (140ms) and not different between age groups. Also asymmetry did not show differences between stepping up or down or between age groups. However, asymmetry revealed that steps with the dominant leg were of equal or faster speed than the non-dominant leg in 43/46 cases ascending and 39/46 cases descending. Stair test parameters were not sign. correlated to other demographic parameters such as gender, height or BMI. Sensitivity and specificity for detecting pathological motion was 0.68 (CI 0.50-0.81) and 0.92 for the most sensitive parameter (step time down). Sensitivity and specificity to detect MT was 0.74 and 0.25 overall compared to 0.53 and 0.50 for the McMurray. Sensitivity increased to 1.00 when MT was combined with a chondropathy scale III or IV (McMurray 0.33). Acceleration based stair climbing test can identify age related degradation in motion, distinguish healthy and pathological subjects and detect the presence of MT with better sensitivity than te classic scores especially when combined with severe chondropathy. REFERENCES [1] Grimm et al. Application feasibility of accelerometer based gait analysis in clinical orthopaedics. J Biomech 39(1), S110, 2006. [2] Van Hemert et al. No. 097, Proc. ORS 2006
Tjouwke van Kalkeren, Eduard van der Houwen
Abstract: ABSTRACT People suffering from high stage throat cancer undergo surgical removal of the larynx as part of their treatment, also called a total laryngectomy. During this procedure the trachea is cut and connected to a hole (tracheostoma) at the base of the neck (Fig 1 and 2). Laryngectomised patients suffer a great number of disabilities due to their altered anatomy. Patients are unable to speak, smell and often suffer from an irritated respiratory tract. Filters and tracheostoma valves attached to the tracheostoma of these patients improve the overall quality of life. Both types of devices need a means that can durably interface them to the stoma. Frequently used for attaching these aids to the tracheostoma are adhesive patches or stoma stickers. Using, applying and maintaining these stickers is perceived to be very cumbersome and disappointing by the patient [1,2]. Most stickers attach very poorly to the skin around the stoma, prone to incompatibility of the shape of the sticker as compared to the skin around the stoma. The aim of this study was to collect worldwide (peri-)tracheostomal geometry data to assess the different shapes and sizes of patients’ stomas. The overall aim is to use the data for future development of a new tracheostomal sticker. This can benefit the daily use of attached valves and filters and as a result a better quality of life. 8 medical centres collected peristomal geometry data of 199 laryngectomy patients. Furthermore, patients completed a questionnaire concerning their experiences with stickers and filters. Data was supplemented using the patient’s medical history. The worldwide obtained data indicates new possibilities for attachment of tracheostoma stickers.
Arjan Knulst, Jenny Dankelman
Abstract: The traditional surgical luminaire is a large, flat-shaped hemisphere, suspended by moveable ceiling arms. The position and orientation of the luminaire are controlled manually. This concept basically did not change during the last 6 decennia, although the introduction of LED technology introduces some deviations from the traditional concept. However, 67% of German surgeons sees room for ergonomic improvements [1]. Also surgeons in other countries complain about lighting [2]. They complain about the experienced quality of surgical lighting during surgery, requiring preoperative manipulations of the luminaire by the surgeon or the assistants. These manipulations not easily performed: ceiling arms collide (78%) or get entangled (63%); one handed manipulations are not always possible (36%) [1].These manipulations take time and concentration from the OR-team, distracting their attention from the patient and delaying the operation. Meanwhile, the large, hot luminaire disturbs the laminar air flow, introducing turbulent air currents from the periphery towards the wound [3, 4]. A system that provides Safe and Adaptive Lighting of the Surgical Area (SALSA) may take over these manipulations and increase patient safety. Measurements of luminaire properties like illuminance, illuminance distribution, irradiance spectra, luminaire temperature were performed at 8 different state-of-the-art (LED) luminaires to gain knowledge on the current performance of surgical luminaires. The measurements were combined with comments from surgeons all using these luminaires to obtain subjective information about the luminaires. LED luminaires had a higher surface temperature than a good halogen luminaire. Not all LED luminaires had an option to focus the light beam. Correlated color temperatures ranged from 2500 to 5300 K. General color rendering indices Ra varied from 88 to 96. The specific color rendering index for red colors R9 varied from 50 to 96. Large diameter luminaires performed better at shadow elimination and had no advantage for deep-wound illumination. The small light field diameters ranged from 200-300 mm, and the large light fields ranged from 290-330 mm. There is a need for improvement of the ergonomics of the surgical light. LEDs provide new possibilities in luminaire design although these possibilities are not fully used yet. Measurements showed that LED luminaires perform equally good or better than traditional halogen luminaires on most luminaire aspects, except for luminaire temperature. The measured data will be used as input for the development of the new SALSA lighting system. REFERENCES 1. Matern, U. and S. Koneczny, Safety, hazards and ergonomics in the operating room. Surgical Endoscopy, 2007. 21(11): p. 1965-1969. 2. Patkin, M., What surgeons want in operating rooms. Minimally Invasive Therapy and Allied Technologies, 2003. 12(6): p. 256-262. 3. Condon, R.E., Nosocomial infection risk from surgical light adjustment. Jama-Journal of the American Medical Association, 1988. 259(20): p. 3052-3052. 4. Hall, G., Air flow disruption must be minimised. Health estate, 2005. 59(3): p. 53-55.
Leo van den Berg, Arnold Schilham, Ruud Verdaasdonk
Abstract: In almost all X-ray systems in the Netherlands, film has been replaced by digital detectors. This change from analog to digital images has resulted in a chain of image processing between the latent X-ray image and the image presented on a viewing station of the radiologist. To understand and predict the resulting image quality, the contrast resolution is assessed at each step in the chain. The first step is after the detector with its analog digital converter, with or without a logarithmic conversion. The image obtained after this step is called the raw image. The second step is the post processing, which includes all image enhancements (e.g. histogram analysis, frequency band processing, look-up-table transformations, noise reduction, etc.). The third step is the reduction of the bit depth to that of the PACS (picture archive and communication system). After this last step, when the image is archived into PACS the contrast resolution is definitive. During any of these steps, information of clinical importance could be filtered out. Since each manufacturer has its own (undisclosed) image processing, there can be major differences in contrast resolution even when the raw image is based on the same type of detector. In this study the clinical performance of digital systems is investigated. For X-ray systems of different manufactures the imaging quality in terms of contrast resolution was determined at each step in the imaging chain for three radiation qualities: RQC3 (50 kV), RQC5 (70 kV) and RQC8 (100 kV). Clinical performance is assessed by investigating the Large Transfer Function (LTF) and the gradient as a function of attenuation instead of radiation intensity with a wedge of aluminum (0 to 50 mm thickness, 250 mm length). To assess the clinical implications of the findings, representative images with a large density range, e.g. lateral lumbar spine, were evaluated by an experienced radiologist. This way, the imaging quality in terms of LTF and gradient was related to the observed image quality. Our results show that the contrast resolution has a high impact on image quality. If the observed image quality is inadequate we are able to pinpoint the weakest link of the processing chain and where the clinical information will be lost. Based on our results we recommend a logarithmic conversion for an adequate image quality. REFERENCES [1] J.T. Dobbins, D.L. Ergun and D.C. Clark, “DQE(f) of four generations of computed radiography acquisition devices”, Med. Phys., 22(10), pp. 158193, (1995). [2] R.S. Saunders and E. Samei, “A method for modifying the image quality parameters of digital radiographic images”, Med. Phys., 30(11): pp 3006-17 (2003) [3] A.D. Maidment, R. Fahrig and M.J. Yaffe, “Dynamic range requirements in digital mammography”, Med. Phys., 20 (6), pp 1621-33 (1993) [4] NEN-EN-IEC 62220-1, “Characteristics of digital X-ray imaging devices. Part 1: Determination of the detective quantum efficiency”, January 2004
Beatrijs van der Hout-van der Jagt, Peter Bovendeerd, Frans van de Vosse, Guid Oei
Abstract: Medical simulation training of obstetric teams is a useful tool in prevention of medical errors. Existing birthing simulators, however, do not have a model-driven system to provide information on the hemodynamic state of mother and fetus; which is necessary for decision making during an emergency. We developed a model-driven circulation model of fetus, mother and placenta to give the obstetric team feedback on the hemodynamic state. The feto-maternal-placental circulation consists of a lumped parameter model based on mass and momentum conservation. The system is developed with DiSCo [1] in Matlab. The maternal and fetal circulation are both driven by a one-fiber model to model the heart. [2] The model is completed with a baroreceptor reflex [3] and an oxygen diffusion model. [4] The model provides heart rate, ejection fraction, stroke volume, blood volume, flow and pressure data. Oxygen pressure is calculated at the fetal and maternal site of the placenta and in the arterial and peripheral fetal circulation. Some of these results include: maternal heart rate (92 bpm), blood pressure (115/50 mmHg), stroke volume (69.8 ml), ejection fraction (0.64); fetal heart rate (142 bpm), blood pressure (80/45 mmHg), stroke volume (6.5 ml) and ejection fraction (0.67). Most output parameters are within target values which are obtained from literature. The model will be further developed and validated with clinical data. Future plans include extension with more clinical relevant parameters, including a pH-model. Once the reference state is fully developed and validated, the model will be extended with obstetric emergency scenarios. REFERENCES [1] Schreurs, P.J.G., http://www.mate.tue.nl/~piet/ [2] Bovendeerd, P.H.M., Borsje, P., Arts, T., van de Vosse, F.N., Dependence of intramyocardial pressure and coronary flow on ventricular loading and contractility: a model study. Ann Biomed Eng 2006;34:1833-45. [3] Van Roon, A.M., Short term cardiovascular effects of mental tasks. Physiology, experiments and computer simulations. PhD thesis, University of Groningen, 1998. [4] Sá Couto, P.M., van Meurs, W.L., Bernardes, J.F., Marques de Sá, J.P., Goodwin, J.A., Mathematical model for educational simulation of oxygen delivery to the fetus. Control Eng Prac 2002; 10:59-66.
Mariken Zijlmans
Abstract: Aim Eye tracking methods are valuable tools in ocular and vestibular testing. In this study was examined how the Maastricht binocular 2D-video eye tracking approach (MVET) performs in the determination of absolute eye position, eye rotation and point of regard, with the main focus on the application in infant subjects. With the advantages of being non-intrusive and not requiring a calibration procedure, MVET could offer an alternative for electro-oculography (EOG), the most widely used method in ocular and vestibular testing in infants. Materials and Methods MVET determines point of regard by means of the position of the virtual images of an infrared light source (the corneal reflexes) and the position of the center of both pupils (appearing as a bright disc) in the grayscale images as recorded by an infrared video camera. Software was developed to determine the absolute positions of the corneal reflexes and the pupil centers in the video images. A model was designed describing the relation of the position of both the corneal reflex and the pupil center with the absolute position and rotation angle of each eye, from which the point of regard can be determined. To examine the actual performance of the eye tracking approach, measurements were performed and analyzed by means of accuracy and sensitivity values. Results Absolute eye position can be determined with a sensitivity of 0.05±0.03 cm in x- and 0.04±0.01 cm in y-direction. Eye rotation can be determined with an accuracy of 0.6±0.4° in x- and 0.3±0.3° in y-direction and a sensitivity of 1.9±0.2° in x- and 2.6±0.3° in y-direction. Point of regard can be determined with an accuracy of 2.1±2.7 cm in x- and 3.2±2.6 cm in y-direction and a sensitivity of 2.4±1.5 cm in x- and 5.2±3.7 cm in y-direction. Compared with the performance of EOG (having a sensitivity of 0.3° and 0.9° in x- and y-direction and an accuracy of >3° in both x- and y-direction), MVET has better accuracy but worse sensitivity in determination of rotation angle of the eyes (for determination of absolute eye position and point of regard, no reference values of the EOG method exist for comparison). The accuracy and sensitivity values are negatively influenced by various factors such as small pupils and non-uniform illumination of the eyes. Conclusion Using MVET in infants has the advantages of not requiring a calibration procedure. Moreover, MVET has better accuracy than EOG. However, the results of this study are obtained under optimal conditions (e.g. well instructable adult subjects, limited head movement). In infant patients the conditions are non-optimal, and the aspects influencing the accuracy and sensitivity values will be even more present. For these reasons, MVET as it is now is not yet a suitable alternative for EOG. Sensitivity could be improved by increasing the resolution of the camera. Then, more measurements in infant subjects could reveal whether MVET does offer a suitable alternative for EOG.
Annemoon Timmerman, Brechtje Riphagen, Ruud Verdaasdonk
Abstract: At Intensive Care units, several drugs are usually administered to a patient simultaneously using syringe pumps. In order to reduce the number of punctures of a patient, the pumps are placed in parallel and the separate infusion flows are passively combined outside the patient using a multi-inlet, single outlet manifold and then delivered into a vein through a catheter. At present, the interference between the lines due to pressure and flow differences is unknown. However, it is suspected and shown in various publications [1,2] that fluctuations in syringe pump infusions can result in uncontrolled and inaccurate administration of medication. Although very few publications can be found [3] that mention the actual problems caused by multi-infusion, this must be equally true for fluctuations caused by the dynamic interactions within multi-pump infusions. To investigate the dynamics and interaction of multi-infusion, a dedicated setup was developed to measure the flow and concentrations of fluids dynamically in multiple lines using absorption spectral-photometry. Through each line, a fluid with a known concentration of a specific absorber streams to the central point were 4 lines merge and the fluids mix. The concentration of each original fluid is determined by a spectral transmission measurement through a flow cell in the central line. The absorbers can be discriminated spectroscopically and the concentration is derived. The flow is measured, collecting the fluid out stream on a weighting scale. The measurements are performed continuously in time. For feasibility testing and calibration, various dyes and concentrations were investigated to find the optimal settings. The method was used for initial experiments and validation for measuring effects of changes in a multi-infusion setup simulating a clinical situation. The method developed for measuring the fluid dynamics of multi-infusion will contribute to a better insight and controlled administration of medication. REFERENCES [1] Fluctuations in syringe-pump infusions: association with blood pressure variations in infants. Am J Health Syst Pharm, Vol 52(15), pp1646-53, (1995) [2] Infusion pump performance with vertical displacement: effect of syringe pump and assembly type Intensive Care Med. Vol 27(1), pp287-91, (2001) [3] T. Rakza, A Richard, A.C. Lelieur, I Villaume, A. Huyghe, C. Nempont and L. Storme, “Facteurs influençant la délivrance de substances perfusées à faible debit par seringue autopulsée: conséquences sur la perfusion de médicaments vasoactifs chez l’enfant prématuré.”, Arch de pédiatrie Vol 12, pp 548-554, (2005).
Marvin Steijaert, Dragan Bosnacki, Huub ten Eikelder, Erik de Vink
Abstract: Genetic information of the DNA is transferred with a very low error rate via mRNA to proteins. In the second phase of this process, translation, an mRNA sequence is processed in the ribosome, leading to the addition of the corresponding amino acids to the nascent polypeptide chain. This process takes place with a very low error rate, which is due to the use of a proofreading mechanism. We have built a simple stochastic model of the translation process and implemented it using the model checker PRISM. With this model we can predict for each codon the expected translation time and the probability of adding a wrong amino acid. The addition of a wrong amino acid for a given codon means that the intended amino acid is replaced by an erroneous amino acid. Hence we can use the error probabilities for individual codons to compute an amino acid substitution matrix. With this matrix we have further investigated the influence of various parameters, including the concentrations of the available tRNA species and their assignment to individual codons. Although 64 codons exist, only a limited number of tRNA species are available (48 in both humans and E.coli). Because of that, many tRNAs have an anti-codon that matches multiple similar codons. In addition, the concentrations of the different tRNA species show a large variation (a factor 60 difference between the most and least expressed species). To study the influence of the available set of tRNAs and their concentrations on the error rates, we have calculated the error rates for number of alternative tRNA sets and alternative concentrations. Apart from the minimization of amino acid substitution, also a second level of error minimization may exist. We hypothesize that substitution by a similar amino acid has generally less deleterious effects than substitutions by a less similar amino acid. Therefore, we have investigated whether substitutions by similar amino acids are more frequent than that of other amino acids. To do that, we have grouped amino acids according to so-called mutation data matrices, like Dayhoff and PET91, which are widely used as an amino acid similarity measure.
Hans van Dijk, Frank van Wamel, Bert Kleine, Dick Stegeman
Abstract: Multichannel or high-density surface EMG (HD-sEMG) provides more extensive and detailed information on a motor unit level compared to conventional surface EMG. A key element in using HDsEMG in a clinical environment is the ability to recognise motor unit (MU) enlargement due to collateral re-innervation in diseases such as Amyotrophic Lateral Sclerosis (ALS). Motor neurons disappear, but the orphaned muscle fibres are innervated by other motor neurons resulting in larger motor units. Motor unit number estimation (MUNE) is a promising tool to monitor such changes1. The MUNE is determined by obtaining a sample of motor unit potentials (MUPs) from a muscle and divide this into the compound muscle action potential (sum of all MUPs). Questions of accuracy and the ability to detect early changes are difficult to address in a patient population without a gold standard available. In the current study we modelled sEMG signals to simulate an ongoing process of denervation and reinnervation like in ALS. We combined two existing models. For surface EMG, a previously described three layer analytical model is used2. The model includes skin, fat, and muscle tissue and can simulate the surface EMG of a single muscle fibre. Furthermore, the model includes the positive wave that occurs when the action potential extinguishes at the muscle fibre-tendon transition, an important aspect when sEMG is measured with respect to a far away reference. To simulate the process of denervation and reinnervation a second model is used that has been described and used to study motor unit potentials (MUPs) with needle EMG3,4. We used the fibre distribution and MU territory information from this model as an input for our surface EMG model. Combining these two models provides the necessary information to simulate the HD-sEMG of a muscle in the process of denervation and reinnervation. We will use the results to estimate the accuracy and sensitivity of MUNE and the sensitivity of surface EMG MUPs to increased MU size and increased MUP instability. Although we focus in this study on the sensitivity of MUNE in detecting changes caused by denervation and reinnervation, the model has greater potential. Since both sEMG and needle EMG signals can be simulated using the above mentioned models, a quantitative comparison between the two methods could be made and the clinical effectiveness of HD-sEMG can be studied as well.
Dedy Wicaksono, Eduardo Margallo-Balbas, Paul Breedveld, Jenny Dankelman
Abstract: Complications during the surgical phase of dental implants still affect a large number of patients. Among these complications are nerve lesions, damage to adjacent teeth, and even life-threatening perioperative bleeding. These problems still occur despite the careful planning phases using advanced technologies, which are expensive, time-consuming, and involve relatively high radiation doses. This paper is a part of the project that aims at radically improves the patient safety in oral implantology, especially during the surgical phase. In contrast with current trends of more demanding planning phase prior to the surgery, the focus of the entire project is on the sensory information available to the practitioner during surgery. To achieve this aim, optical analysis of the tissue surrounding the implantation site will be performed using fibre optics integrated into the drill bit. A rotating optical coupler will be developed to guide light between the hand-piece and the rotating drill bit. Optical Coherence Tomography (OCT) will be used as a method for structural analysis of the tissue by processing the interference pattern coming from the fibre optics. The principle of the device used to process this optical signal is reported elsewhere by one of the authors (EMB) [1]. This particular paper reports the mechanical design of the integration between the sensor probe, which consists of the fibre optics and the gradient index (GRIN) lens, and the drill bit. The fibre optics and the GRIN lens need to be integrated within a 2-mm diameter drill bit. To proof the principle, a steel probe resembling a real drill bit is designed. The probe will consist of three different parts: the cone-shape tip, the middle tube, and the hook part. The cone-shape tip is the part that will be containing the GRIN lens with a focal of 1 mm, for scanning the surrounding tissue. The fibre optics will be glued and attached to the GRIN lens inside this cone-shape tip. The fibre optics will stretch along inside the hollow part of the middle tube. The other end of the fibre will be glued to another GRIN lens, with a focal length of infinity, inside the hook part. This hook part will be screwed to the hand-piece of the drilling system. Several possible design alternatives will be presented along with each advantages and disadvantages. REFERENCES [1] E. Margallo-Balbás, G. Pandraud, and P.J. French, “Miniature Optical Coherence Tomography System Based on Silicon Photonics”, Proc. SPIE, Vol. 6847, pp. 68470S, (1993).
Tjitske Anke Boonstra, Edwin van Asseldonk, Jeroen van Vugt, Herman van der Kooij, Bastiaan Bloem
Abstract: Parkinson’s disease (PD) is a progressive, incapacitating neurological disease negatively affecting the quality of life for many reasons, not the least because of severe motor impairments. Symptoms include appendicular disability, namely rigidity, bradykinesia and tremor, but also axial disability: gait disorders, balance impairment, falls and fall-related injuries (postural instability and gait disorder, PIGD). Parkinson’s disease is an asymmetrical disease; symptoms usually start on one side of the body and throughout the course of the disease this side remains the most impaired. Asymmetry of symptoms is clinically evident for bradykinesia, rigidity and tremor. However, it is uncertain whether asymmetries also underlie the PIGD. Clinical scales used to assess disease severity (e.g., the Unified Parkinson’s Disease Rating Scale (UPDRS)) do not explicitly evaluate asymmetries in axial symptoms. Evaluation of such asymmetry might improve our understanding and treatment of PIGD In this study, a new method developed by Van der Kooij and colleagues [1, 2] was used to test the hypothesis that balance control in PD is asymmetrically affected. Eight patients with idiopathic Parkinson’s disease were asked to maintain their balance without moving their feet, while standing still and during continuous random platform movements, with eyes open (EO) or eyes closed (EC). Body sway angle, reactive forces of each foot and EMG of the lower leg muscles were recorded. These measurements yielded the Frequency Response Function (FRF) of the stabilizing mechanisms, which expresses the amount and timing of the generated corrective torque in response to sway at the specified frequencies. The FRFs were used to calculate the relative contribution of each ankle to the total amount of generated corrective torque to resist the perturbations. In addition, the motor part of the UPDRS and the Hoehn & Yahr disease stage were also evaluated. Preliminary results showed that all patients exhibited abnormal asymmetrical dynamic weight bearing: one leg was bearing more weight than the other leg. In addition, responses to platform perturbations as shown by corrective ankle torques were also asymmetrical, even more so than dynamic weight bearing. Hence, one leg contributed significantly more to balance control than the other and there was no clear relationship between the contribution to weight bearing and to balance control. In contrast, a group of healthy subjects instructed to distribute their weight asymmetrically showed a one-on-one relation between weight bearing and balance [1]. In summary, balance control in PD proved highly asymmetrical. This asymmetry of axial symptoms can be reliably identified with system identification techniques in the frequency domain. In this way, the effects of different treatments such as drug therapy, physiotherapy and deep brain stimulation can be identified for each leg separately. This will provide a better understanding of the pathophysiology of PIGD disease and the effects of different treatments. 1. van Asseldonk, E.H., et al., Disentangling the contribution of the paretic and non-paretic ankle to balance control in stroke patients. Exp Neurol, 2006. 201(2): p. 441-51. 2. van der Kooij, H., et al., Detecting asymmetries in balance control with system identification: first experimental results from Parkinson patients. J Neural Transm, 2007. 114(10): p. 1333-7.
Loubna Bouarfa, Jenny Dankelman
Abstract: In the last decade, many new technologies have entered the operating room (OR). Although highly sophisticated technology is used within the OR, software to aid managing workflow in the OR is poorly developed compared to other industries. In conventional industries, workflow systems are used to manage the work inside the company. Workflow may be seen as an abstraction or virtual representation of actual work and calls upon the appropriate tools. It supplies necessary information, business rules, checklists, task lists and reports to the Employee. Employing workflow systems to automate high throughput, repetitive processes, such as checklist reviews, can have profound effect on every aspect of the process. There are several benefits for the work environment of any organization on the implementation of workflow principles. Studies [1] have shown that implementing the workflow system in a business process results in an enhancement of business operations. In the operating room, there are high-level processes carried out by the surgical team. To describe and record the activities of human beings, we need very high level descriptors. One may object to this line of argument by remarking that computers are routinely used in industrial applications (i.e. workflow systems) in which human beings and situations concerning them are involved and that relevant computations are carried out successfully. However, in conventional industries, the description of a human being have been so highly abstracted that the events or processes have been reduced to low level objects. For example in the banking industry the human being is abstracted as a customer Alice and a banker Bob. To describe commercial activities like internet banking, we need only a few properties; both the customer (Alice) and the banker (Bob) are abstract entities "Person A wants to get a loan form person B”. In the operating room, such abstraction is Impossible [2]. This is caused by the more complex behaviors that human beings display in the OR (including aspects like human cognition, self-consciousness, and behavior). In this case, one must deal with the rich description occurring at higher level of behavior. The workflow in the OR is dependent on the dynamic information flow between the patient anatomy, perception, cognition of the surgical team, and the state of the machines (monitors, equipment and surgical instruments). This dynamic interaction requires the design of dynamic expert systems that can understand, learn and guide the workflow in the OR instead of enforcing a linear workflow [3]. The goal of my PhD research is to design a workflow expert system suitable for the OR environment; the Digital OR Assistant (DORA). Its design goal is to log a specific surgical procedure and guide the reasoning of the surgical team at an expert level. Allowing for safer and more efficient surgical procedures.
Jaap Harlaar
Abstract: This study evaluates the accuracy and the precision of a newly developed Femoral Epicondylar Frame to measure transversal plane kinematics of the knee in patient with a knee prosthesis. Osteoarthritis is an age-related degenerative joint disease and affects a substantial part of the population over the age of 50, and is successfully treated with a knee endo-prosthesis. However, the behaviour of the (mobile bearing) prosthesis in the transversal plane during daily tasks is still under debate. Interpretation of optoelectronic recordings of skin mounted markers are hampered by skin artefacts. Therefore we designed the Femoral Epicondylar Frame (FEF), using clamping pressure on custom made epicondylar attachments. In order to evaluate the precision and accuracy of the FEF in the patient target population, we used fluoroscopic stereometry as a golden standard, using the TKA components as bone reference, during a step-up task. Fluoroscopic stereometry is accurate within 0.3° [1] Seven patients (mean age 71 years), with a BMI ranging from 26 to 30 were measured. All subjects had received total knee replacement three to four years prior to the study. The patients were asked to perform three step-up tasks, with the knee centered between the image intensifier and focus of a fluoroscope [1]. A three dimensional model of the TKA was fitted to the planar image to reconstruct the in vivo position and orientation of the the TKA as well as the FEF markers [1] No significant differences were found for knee flexion angles below 40°, indicating a good accuracy. RMS error was linearly dependent on the flexion. For knee flexion angles below 40° the precision is always better than 3.3°, while its mean is less than 2.2° over the whole range of knee flexion, i.e. 0°-40°. The amount of error we found was similar to another study of Südhoff et al. [2], who used a rather painful clamping technique. Houck at al [3] used a simpler clamp, but found an error of more than 5° within the 0-40° knee flexion range. Moreover, we used a group of typical subjects with higher BMI’s, so it can be concluded that the FEF is performing superior over existing femur attachments sytems. However, the resulting precision , i.e 3.3°, is in the same order of magnitude as the differnces that might observed in comparing differrent knee endoprosthesis. Idellay one would like a better presicion by one order of manitude, like fluorscopy [1]. However, optoelectronic methods are much more feasible to be used for different tasks, e.g. turning away while rising from a chair, to provoke knee axial rotation [4]. Moreover optoelectonic methods are non-invasive. The precision found will not allow to be conclusive on an individual level, but requires multiple measurements in a group of patients. Therefore it is concluded that use of the FEF with optoelectronic stereometry is a valid method to track axial femoral rotation for most clinical studies [1] Garling, E.H., et. Al 2005. Journal of Biomechanics 38, 893-901 [2] Südhoff, I., et al. 2007.. Gait & Posture 25, 533-543. [3] Houck, J., 2004 Gait & Posture 19, 76-84. [4] Zürcher AW, et al. 2008 Gait & Posture 28(3): 472-477
Magdalena Chmarra, Stefan Klein, Joost de Winter, Frank-Willem Jansen, Jenny Dankelman
Abstract: Background: From the clinical point of view, it is important to know which group – expert, intermediate, or novice – the resident belongs to according to his/her basic psychomotor skills. For that reason, surgeons and surgical organizations (e.g., ACGME) are calling for assessment tools that credential residents as technically competent. Currently, no method is universally accepted or recommended to classify residents as experts, intermediates, and novices according to their technical skills. This study introduces such a classification method in laparoscopic surgery. Methods: 10 experts (experience > 100 laparoscopic procedures), 10 intermediates (10-100 procedures), and 11 novices (no experience) performed four tasks in a box trainer. Movements of laparoscopic instruments were recorded with the TrEndo tracking system and analyzed using six motion analysis parameters (MAPs). The MAPs of all participants were submitted to Principal Component Analysis (PCA), a data reduction technique. The scores of the first principal components were used to perform Linear Discriminant Analysis (LDA), a classification method. Performance of the LDA was examined using a leave-one-out cross-validation. Results: 23 (74%) participants were correctly classified with the proposed method: seven experts, seven intermediates, and nine novices. Conclusions: The proposed method provides a means to objectively classify residents as experts, intermediates, and novices according to their basic laparoscopic skills. Because of the simplicity and generalizability of the introduced classification method it is easy to implement it in existing trainers.
Jurriaan Bakker, Maarten Paulides, Tomas Drizdal, Richard Canters, Gerard van Rhoon
Abstract: Hyperthermia (HT) is considered a valuable anti-cancer treatment modality in several Western countries. For advanced cervical cancer adding HT to radiotherapy (RT) doubles the 3 years survival rate (27 vs. 51%). For recurrent breast cancer in earlier irradiated areas the local control rate for RT+HT is doubled compared to RT alone (39 vs. 79%). Generally, HT is applied by electromagnetic (EM) radiation using complex equipment. A strong relation exists between treatment quality and clinical outcome; hence controlled delivery of heat is the Achilles heel of the treatment. Current efforts to enhance the quality are focusing on our ability to improve dosimetry and to prospectively prescribe thermal dose. An absolute requirement for future solutions is that they must be economically as well as clinically acceptable. Adequate monitoring of HT treatment quality is a very difficult task. Neither the patient nor the clinician appreciates extensive interstitial thermometry. Non Invasive Thermometry (NIT) by Magnetic Resonance Imaging (MRI) is the most recent technological improvement, however not the solution for all HT applications. The tremendous improvements in the available software for 3D EM modeling offer unique possibilities to: · Optimize the quality of the HT treatment prospectively or during treatment. · Establish new Quality Assurance and control guidelines. · Guide selection of optimized (position, amplitudes & phases) applicator settings · Select the most promising applicator. Hyperthermia Treatment Planning (HTP) may potentially increase the tumor. Most recently our group exploited EM modeling for the development of a complex array applicator for HT treatment of head & neck tumors. We are convinced that these efforts will improve clinical outcome and reduce toxicity
Ferry Melchels, Jan Feijen, Dirk Grijpma
Abstract: In tissue engineering, porous scaffolds are used as a temporal support for tissue regeneration. We use stereolithography as a manufacturing tool and a resin based on poly(D,L-lactide), to make rigid biodegradable scaffolds having a well-defined architecture. Our objective was to investigate the effect of scaffold architecture on the mechanical properties. Different architectures were built and tested under compression, comparing the results to predictions obtained by mathematical modelling.
Eka Budiarto, Marleen Keijzer, Pascal Storchi, Arnold Heemink, Ben Heijmen
Abstract: Radiation therapy is used to fight cancer by giving a lethal radiation dose to the tumour. Unfortunately, this radiation can also damage the healthy organs around the tumour. With Intensity Modulated Radiation Therapy (IMRT), not only the angle of radiation beams, but also the intensity profile of the beams can be adjusted, making the treatment more accurate. All requirements for the radiation therapy, for example not giving too high a dose to healthy organs and giving a sufficient dose to the tumour, are translated into constraints, making the problem in general a multi-criteria optimization. Nevertheless, radiation planning is based on static geometrical information of the organs (CT-scans), which might not give an accurate description due to random movements and deformations of the organs, both within and between radiation fractions. They can be caused for example by respiration, gas flowing through the colon, or by inaccuracies in the patient position on the treatment table. All of these create geometrical uncertainties for the radiation planning. To cope with these uncertainties, the radiation area is usually made larger than the tumour itself by adding some margins around it. This way, the tumour will always get a sufficient dose in spite of the movements and deformations. On the other hand, this method will also damage more healthy tissue around the tumour than necessary. Therefore, a model to characterize the random movements and deformations, as well as an optimization process based on this model are needed to make the margins as small as possible. Previous results show that a model based on principal component analysis (PCA) [2] can be used to characterize the random movements and deformations into their dominant modes, reducing the number of parameters to a manageable size. This method needs a lot of CT-scans, which are usually not available for one particular patient before treatment. A method to do an optimal PCA based on data from a representative patient population is currently under development. The result of this PCA will be implemented into an optimization algorithm [1] to compute the intensity profile of radiation beams given angles of the beams. This algorithm, developed in our departments, has been shown to give good and realizable radiation plans in the static case (i.e. without geometrical uncertainties). The necessary modifications will be made to explicitly include the geometrical uncertainties. REFERENCES [1]S. Breedveld, P.R. Storchi, M. Keijzer, A.W. Heemink, and B.J. Heijmen. A novel approach to multi-criteria inverse planning for IMRT. Phys. Med. Biol., 52:6339-6353, 2007 [2]M. Soehn, M. Birkner, D. Yan, and M. Alber. Modelling individual geometric variation based on dominant eigenmodes of organ deformation: implementation and evaluation. Phys. Med. Biol., 50:5893-5908, 2005
John van den Dobbelsteen, Ivo Schoemaker, Jenny Dankelman
Abstract: Purpose Simulation systems may offer an environment where competence can be acquired in the precise positioning of catheters while patient safety is not at risk. At the proximal end of the catheter the interventionist experiences resistive forces resulting from contact between the catheter and the introducer sheath and forces due to contact between the instrument and the vessel wall. The goal of the current study is to determine the component of friction forces during propagation of a catheter for the development of simulation models. Materials and methods Measurements were performed in two pigs using a 5F and a 6F guiding catheter in combination with a 9F introducer sheath. The catheter was attached to a force sensor and guided into the aorta from the femoral. A motion capture system was used to record the movements of the catheter. The insertion length of the catheter was varied to determine the friction caused by contact between the catheter and the vessel wall. The resistive forces thus found were compared with those caused solely by the sheath. Results The resistive force was on average 1.2 N (range 0.99 N – 1.44 N). The forces reduced when the insertion length increased. The friction generated at the introducer sheath was 0.63 N and 0.51 N for the 6F and 5F catheter, respectively. The combined results show that the contribution of forces due to contact with the vessel wall ranged between 0.30 and 0.68 N for the 6F catheter and between 0.66 and 1.21 N for the 5F catheter. Conclusions Friction at the introducer sheath can be as high as 0.6 N and the resistive forces are increased by contact between the catheter and the vessel wall. It is likely that most of the friction was caused at the tip which is pushed against the vessel wall due to its curvature, and that this effect decreased because the diameter of the vessel increases at increasing insertion lengths. We conclude that the main determinants of the resistive forces during propagation of a catheter are friction generated at the introducer sheath and friction generated at the tip of the instrument.