6th Dutch Bio-Medical Engineering Conference
26 & 27 January 2017, Egmond aan Zee, The Netherlands
15:00   Poster session I
Emiel van Disseldorp, Julia van Dronkelaar, Josien Pluim, Frans van de Vosse, Marc van Sambeek, Richard Lopata
Abstract: Abdominal aortic aneurysms (AAAs) can lead to a fatal haemorrhage when ruptured. To predict the rupture risk of an AAA, Computed Tomography (CT) based wall stress analysis has been proposed in literature. CT is also considered the gold standard imaging technique to determine the 3D AAA geometry. However, it has some drawbacks, e.g., ionising radiation exposure. As an alternative, 3D ultrasound (US) has shown its feasibility to determine AAA geometry and wall stress. The major limitations were the lack of automated segmentation and the suboptimal field-of-view. Therefore, this study aims to assess AAA geometry based on multiple 3D datasets using automatic segmentation and registration. Results are validated with CT. Thirteen patients with an AAA diameter of 35 – 58 mm were included prospectively. For 8 patients the field-of-view of one 3D US acquisition was insufficient to capture the complete AAA geometry. Therefore, a proximal and distal sub-volume were acquired. For 7 patients US exams were performed at several instances, which resulted in a total of 21 AAA geometries. For each patient CT data were available. Based on an elaborate filter analysis, a 3D Euclidian shortening flow filter was applied to emphasize the lumen- wall interface. Subsequently, a 2D active contour model was applied slice-by-slice to segment the 3D AAA geometry, regularized by a 3D active contour. For the 10 proximal and distal sub-volumes registration was applied using a modified Powell search. Rigid translation and rotation were estimated by maximising the similarity index (SI) between the two segmented volumes. Finally, the image data were merged after which the complete AAA geometry was re-segmented. All AAA geometries were validated by comparing the results to CT-based contours quantitatively using the SI and Hausdorff distance (HD). The single and merged US-based geometries showed a good agreement with the CT results (SI = 0.78 – 0.93) The HD values were higher (median = 7.0 mm) at the proximal and distal sides compared to the middle (median = 5.4 mm). Moreover, the median HD decreased by 23% when registration and re-segmentation were applied. This study shows that automatic segmentation and registration of multiple 3D US volumes of AAAs is feasible to determine the AAA geometry using 3D US. In future studies, the use of image feature-based registration will be investigated and work on AAA wall stress analysis will be extended.
C.C. Roossien, C.T.M. Baten, J. Stegenga, M.F. Reneman, Bart Verkerke
Abstract: Aging of employees comes with physical limitations, including a decrease in muscle strength. This will result in employees having difficulties performing physical active jobs and a misbalance between their individual capacity and workload. A sensor system of inertial measurement units (IMUs) and surface electromyography (sEMG) may be suitable to objectively monitor the working posture and net moments in the lower back in lab conditions [1]. Aims of this study were (1) to investigate if this system can reliably monitor the working posture, compared to Free Body Diagram (FBD); (2) to validate the back load calculation; (3) to investigate is this system is usable in a working situation. Subjects were 19 volunteers (8 females, 11 males), age 20 to 67 years (35.8±14.0 years). The subjects preformed a set of arm and trunk movements and a lifting task. The arm and trunk movements contained standing in a static position for 10 and 35 s and dynamic movements which were performed 10 times. During the lifting task, the subjects lifted a weight of 1.8, 6.3 and 10.3 kg from the floor to a table. Six wired IMUs, (Xsens MTx) were placed on the subject; upper and lower arms, sternum and pelvis. The data analysis software was developed by Baten et al. [2] [3] [4]. sEMG was analyzed with the Polybench Dipha (Inbiolab, Roden, The Netherlands). The electrodes were placed on the longissimus thoracis muscle at L1, the iliocostalis lumborum muscle at L2-L3 and multifidus muscle at L5 with a reference electrode placed at the processus spinosus of C7. Per muscle two unipolar electrodes are placed to measure bipolar. Video recordings were made for calculating the FBD. This lab study is recently finished and the data will be presented at the congress. REFERENCES [1] Kingma, I. et al., „Lumbar loading during lifting: a comparative study of three measurement techniques”. Electromyogr. Kinesiol. 11(5):337-45, 2001. [2] Baten, C.T.M., et al., „Calibration of Low back load estimation through surface EMG signals with the use of artificial neural network technology”. Proceedings of the IEEE Eng. in Med. & Biol. Soc., 17th Annual Int. Conf. Montreal, Quebec, Canada, 1995. [3] Baten, C.T.M., „Ambulatory low back load exposure estimation”. 14Th Triennial Congres of the International Ergonomics Association. San Diego, USA, 2000. [4] Baten, C.T.M., et al., „3D ambulatory motion analysis; already usable in practical (industrial) physical therapy?” (Dutch language). Special Issue Wetensch. Tijdsch. v. Fysioth, 117(6) p233-241, 2007.
Kostas Nizamis, Sander Voss, Bart Koopman
Abstract: Duchenne Muscular Dystrophy (DMD) is a progressive muscular disease. Active hand orthoses can greatly improve the quality of life of people with DMD. One way to control such devices is via forearm sEMG control. Normally sEMG from the forearm is mapped to a gesture in a physiologically relevant way (intuitive). Often, in complex orthoses with many degrees-of-freedom, this imposes extra burden on the user. This research aims to compare the performance of non-intuitive control against intuitive for future use on the control of hand orthoses for people with DMD. We hypothesize that non-intuitive classification can achieve better results and reduce the burden on the user. The experiment was designed in such a way, that difficult gestures for the classifier were mapped to simpler gestures. The easy gestures are: extend wrist, flex wrist, close hand and supination, while the difficult grips are side grip, fine grip, point and agree. Seven healthy subjects were asked to control a virtual hand using both a non-intuitively trained classifier and an intuitively trained. The performance was evaluated based on completion time and completion rate of gestures. It was found that non-intuitive control performed better on completion time and completion rate (p<0.001), however this statement is supported by a limited number of subjects and trials. The subjects also showed a general preference towards the non-intuitive control via a questionnaire. We conclude that non-intuitive control has a great potential for simplifying the control of active hand orthoses and reduce the burden on the user.
Kostas Nizamis, Maarten Ganseij, Bart Koopman
Abstract: Duchenne Muscular Dystrophy (DMD) is a progressive muscular disease. Active hand orthoses can greatly improve the quality of life of people with DMD. Surface Electromyography (sEMG) is commonly used for the control of active devices. The interfacing between the human and the sensor is regularly done by an adhesive skin interface (sticker). This can cause discomfort, especially during daily use. For forearm sEMG measurements, a sleeve design can solve this problem. The design presented here aims to make sEMG more comfortable, yet functional for daily use. In order to achieve that, we designed a simple, low-cost sEMG sleeve using a commercial ankle brace. Six equidistant cuts around the circumference of the sleeve were made in order for the sEMG sensors (Delsys Trigno, Delsys Inc.) to be placed. Those are held in place by a number of 3D printed plastic casings mounted with plastic snap buttons. The buttons are used to make the attachment of the casings fast and easy. A preliminary evaluation of the sleeve has been carried out with six healthy subjects, using a library of 6 and 9 gestures and a simple artificial neural network (ANN) classifier. The performance was evaluated, in terms of classification time, training time and accuracy (offline) and selection time, completion time, completion rate and accuracy (online). The results indicate that the performance of the sleeve is not significantly different than the adhesive skin interface in any metric (p<0.01). We conclude, that the EMG-sleeve is a better alternative than the current commonplace methods for sensor skin interfacing, while having similar performance during a classification task.
Walid Abdelmoula, Benjamin Balluff, Sonja Englert, Jouke Dijkstra, Marcel Reinders, Axel Walch, Liam McDonnell, Boudewijn Lelieveldt
Abstract: Mass spectrometry imaging (MSI) is a technology that simultaneously provides the spatial distribution of hundreds of biomolecules directly from tissue. The identification of tumor subpopulations that adversely affect patient outcome is essential for a more targeted investigation of how tumors develop detrimental phenotypes, as well as for personalized therapy. MSI has demonstrated its ability to uncover molecular intratumor heterogeneity. The challenge has been an objective analysis of the resulting data to identify those tumor subpopulations that affect patient outcome. Here [1] we introduce spatially mapped t-SNE, a non-linear visualization of the data that is able to better resolve the biomolecular intratumor heterogeneity; in an unbiased manner, it uncovers tumor subpopulations that are statistically linked to patient survival in gastric cancer and metastasis status in primary tumors of breast cancer. MALDI MSI datasets of the tumor regions from patient series of gastric cancer (n=62) and breast cancer (n = 32) were analyzed by 3D t-SNE. The images created by t-SNE revealed molecularly distinct subpopulations. An edge-detector was used to highlight the transitions between those distinct subpopulations. To find a discrete representation, edge correlations between t-SNE and k-clustered images were calculated. The clinical relevance of the identified subpopulations was investigated through association with patient survival (gastric cancer) and metastasis status (breast cancer). A statistical analysis revealed several tumor subpopulations were associated with the tumor’s phenotype, and distinct biomarkers identified. The stability of the algorithm to identify phenotypic subpopulations was investigated and the prediction capabilities were then demonstrated by creating a KNN classifier and performing an extensive unbiased cross-validation. An unsupervised, and data-driven, method is presented to identify clinically-relevant tumor subpopulations that drive patient outcome. For more detailed information we refer the interested reader to [1]. REFERENCES [1] W. M. Abdelmoula, B. Balluff et al, “Data Driven Identification of Prognostic Tumor Subpopulations Using Spatially Mapped t-SNE of Mass Spectrometry Imaging Data”, PNAS (in press).
Alessandro Urso, Marijn Van Dongen, Wouter Serdijn
Abstract: This work presents the design and the results from both schematic and post-layout simulations of a multichannel neural stimulator. The neurostimulator uses a different way of stimulating the neural tissue compared with state of the art constant-current stimulators. In fact, each stimulation phase is made of a sequence of current pulses injected at a frequency of 1 MHz, in which a duty cycle signal is used to control the stimulation intensity [1]. Significant size reduction is achieved by using only one inductor as external component. In contrast, conventional constant-current stimulators that use a DC-DC converter as a core, require at least two external components, both an inductor and a capacitor. The improvements with respect to [1] target to increase the power efficiency. The external high-voltage supply, usually used in current-steered stimulators, is avoided. Moreover, the employed zero-current-detection technique together with a power-efficient way of driving the high-voltage switches allows to achieve a record power efficiency while stimulating 8 channels connected to 16 reconfigurable electrodes in a time-interleaved fashion. The circuit is implemented in a 0.18 μm HV process, and the total chip area is 3.6 mm 2 . Post-layout circuit simulations confirm the correct operation: when all the channels are used simultaneously and the load is modeled as a resistance R load = 600 Ω with a series capacitance C load = 500 nF, the power efficiency is increased to 70%. Because of the unique operating mechanism, [1] is the most suitable work for a power-efficiency comparison. When all the channels are used simultaneously, the power efficiency of this work is improved by up to 200% when compared to [1].
Stefan Remmers, Keita Ito, Sandra Hofmann
Abstract: With the current knowledge on bone cells and intercellular communication, we are still not able to reverse bone diseases such as osteoporosis but merely slow the progression of the disease. Recently, it has been suggested that osteoclasts activity is dependent on blood oxygen content, and this may play a role in osteoporosis1. In vitro models have been designed to study the processes underlying bone remodeling, but commonly use animal cells or immortalized cell lines, usually in 2D culture. However, human primary cells are believed to be more biologically relevant than cell-lines or animal cells2, as are 3D cultures when compared to 2D cultures3. We aim to develop a biologically relevant 3D in vitro system in which human primary osteoclasts are cultured on mineralized constructs where resorption is monitored non-destructively over time that can be used for fundamental research on bone resorption. Recently, we managed to differentiate human primary monocytes towards osteoclasts in 2D. We were able to clearly distinguish between monocytes and osteoclasts morphologically as well as by their resorptive capabilities on polished bone and dentin discs as others have previously4. Currently, we are investigating whether Wnt stimulation or blocking during culture can affect osteoclastic resorption, and whether or not this can be quantified into resorption pit number and average diameter. If so, then resorbed mineralized tissue volume can be used as a marker for osteoclastic activity that can be monitored over time in 3D experiments. Then, osteoclasts will be cultured on 3D mineralized constructs and resorption will be monitored over time using micro-computed tomography. This will allow the localization and quantification of osteoclastic resorption within the same sample over time. Finally, the system will be used to study the effect of pericellular oxygen content on osteoclastic resorptive activity. If blood oxygen content indeed affects osteoclastic resorption, then making an extra effort to ensure proper oxygen supply to bones at risk could delay the onset of osteoporosis, or perhaps even prevent it or reverse it.
Ronald Bos, Dick Plettenburg, Just Herder
Abstract: Hand exoskeletons, hand rehabilitation robots and robotic gloves are all examples of names given to dynamic hand orthoses. They can be designed for different patient groups, functionalities and working environments, hence the current state-of-the-art knows a large amount of devices and a high diversity in the used mechatronic components. However, there is also an unbalanced distribution in use of these components due to a focus on mainly highly prevalent patient groups (e.g. stroke) and traditional components (e.g. low-level state control, DC motors, cable transmissions). For more specific patient groups, like boys with Duchenne Muscular Dystrophy (DMD), this asks for a more thorough analysis of the possible solutions. The goal of this study is to provide novel tools and technologies for the design of dynamic hand orthoses, with the intention to find solutions suitable for DMD patients. This includes the development of a structured overview of the possible solutions (categorization), further investigation of outliers (scoping) and optimization of chosen methods (optimization). A categorization based on 165 unique dynamic hand orthoses has revealed 73 possible mechatronic components, which were all categorized in a signal, energy and mechanical domain. A further scope of these components revealed that using compliant mechanisms and miniature fluidic actuation appears most suitable for a dynamic hand orthosis for DMD patients. Subsequently, using tools from engineering optimization, an analysis of several topologies and shapes that incorporate these components has shown to create various possible conceptual solutions. The results show that novel solutions for a dynamic hand orthosis can be reached with a broad approach. Depending on the bounding design criteria imposed by the patient group, functionality and working environment, different conceptual solutions can emerge. This generalized procedure allows it to be used by other researchers from similar research areas as well. The work is part of the Symbionics projects on the hand, which are supported by STW (NL, #13524 and #13525), Hankamp Rehabilitation (Enschede, NL), Hocoma (Volketswil, CH), TMSi (Oldenzaal, NL), FESTO (Delft, NL), Pontes Medical (NL), Spieren voor Spieren (NL) and Duchenne Parent Project (NL).
Laura van Huizen, Nikolay Kuzmin, Marie Louise Groot, Ellis Barbé, Susanne van der Velde, Lisette te Velde
Abstract: ABSTRACT The most common cancer among women is breast cancer. Fast diagnosis and precise excision of the breast tumor is important. Both breast biopsies for diagnosis and breast surgery for treatment are performed in the operation room, after which the excised breast tissues are assessed by a pathologist. At this moment, the standard procedure is formalin-fixation, paraffin-embedding, sectioning, and staining with hematoxylin and eosin (H&E) staining [1]. Disadvantages of this procedure are that this process takes a long time, on average 16 hours, and it utilizes dyes that can damage the tissue [2-4]. Alternative techniques are therefore required that can assess breast tissue without dyes and with a speed that enables ‘live’ feedback to the surgeon while she/he operates. A technique that meets these requirements is higher harmonic generation microscopy (HHGM), a novel imaging technique, which is non-invasive, label-free and provides 3D images with a high, sub-cellular resolution, within seconds [5]. HHGM requires no external contrast agents and reveals tissue contrast provided by interfaces, noncentrosymmetric molecular structures or autofluorescent organelles via generation of second and third optical harmonics (SHG/THG) and autofluorescence. 3D images were recorded without any preparation of healthy human and mouse breast tissue. The images showed that with the combination of THG and SHG microscopy breast components – lobules, ducts, fat tissue, connective tissue, blood vessels, and peripheral nerves – can be identified. All THG/SHG images are in good agreement with H&E histology. Moreover, with a DNA staining experiment we proved that THG microscopy is able to reveal cells, which is necessary to identify lobules and ducts. Furthermore, the experiments showed that HHGM is able to reveal most of the pathological breast features. Thus, HHGM has a high added value for fast label-free assessment of fresh breast tumor tissue. One of the applications of HHGM is in vivo imaging with a bioptic needle. First characterization experiments with a needle shaped Bioptic Needle micro-objective showed that this needle is suitable for THG and SHG imaging. This enables in-situ optical biopsies in the future, which makes HHGM an even more promising technique for the clinical assessment of fresh breast tumor tissue. REFERENCES [1] N.V. Kuzmin, P. Wesseling, P.C.d.W. Hamer, D.P. Noske, G.D. Galgano, H.D. Mansvelder, J.C. Baayen, and M.L. Groot, “Third harmonic generation imaging for fast, label-free pathology of human brain tumors”, Biomedical Optics Express, Vol. 7, pp. 1889–1904, 2016.
Bob Giesberts, Marieke van der Steen, Edsko Hekman, Patrick Maathuis, Bart Verkerke
Abstract: Clubfoot (talipes equinovarus) is a common congenital deformity and often treated directly after birth with the Ponseti method [1]. This treatment method consists of serial manipulation and casting of the deformed foot. This manipulation stretches the tissues on the medial side of the foot and is maintained for a week with a plaster cast. Most cases of clubfoot are corrected after six cast changes and, in many cases, a percutaneous Achilles tenotomy. After the casting period a foot abduction brace is used for four years to prevent relapse. There are good logistical and organisational reasons to change the cast on a weekly basis, and we are looking for possible biomechanical reasons that support the cast to be left on for a full week. We performed a literature review on accelerated versions of the Ponseti method in which the cast was changed more than once a week. There is strong evidence that changing the cast twice per week is just as effective without increasing complications or the required number of casts [2-4]. By doing so, the duration of the casting period of the treatment is decreased from six to three weeks. This suggests that from a biomechanical point of view there is no reason to leave the cast for a full week. To further investigate the biomechanics behind the Ponseti method we are currently performing a new study in which we measure in ten subjects how the cast/skin-interface pressure decreases over time as a result of tissue adaptation in the clubfoot. We expect that the measured force will decrease rapidly in the first few hours and reach an equilibrium within the first day. At time of writing the first sensors are placed underneath the cast of the first clubfoot patient. Preliminary results will be presented at the Dutch BME2017.
Nikita Kruis, Marica Emmer, Michaël Lansbergen
Abstract: Quality and safe use of medical technology should be assured by multidisciplinary life cycle management. [1] However, multidisciplinary collaboration in hospitals is recently introduced and due to missing overview of the state of health care technology infrastructure, the quality and safe use of the medical equipment is not always optimally assured. [2][3] The need for a multidisciplinary dashboard with clear information of the state of medical technology infrastructure has only grown since the government introduced the covenant “safe use of medical technology in hospitals”. [4][5] Our aim is to determine which information about medical equipment is desirable, the best way to present such information and who should have access to it. An audit was performed in the Rijnstate hospitals. Possible stakeholders were interviewed, processes used in practise were compared to the documented processes, and registered information of the medical equipment was gathered. This was used to design information structures and explore technical dashboard designs. Results so far showed that a lot more stakeholders are involved than the organization assumed. These stakeholders indicate a need for clear information. Legal regulations are implemented; however, in daily practice assurance is not always sufficient and registration of medical equipment is not uniform. In this project we created multidisciplinary support, determined the desired information and designed information structures. In addition we found that a multidisciplinary designed dashboard providing a clear overview of information about the state of medical technology and regulatory considerations of medical equipment is valuable to a large variety of stakeholders in a healthcare organization. Implementation of such a dashboard will encourage multidisciplinary collaboration and improve quality and safety assurance in health care technology lifecycle management. REFERENCES [1] R.J. den Adel and M.D.I Lansbergen, “Complexiteit de baas?! Zorgtechnologie is mensenwerk”, Technische Universiteit Eindhoven, (2014). [2] IGZ, Veilig gebruik van medische technologie krijgt onvoldoende bestuurlijke aandacht in ziekenhuizen, 2014 [3] IGZ, Staat van de gezondheidszorg 2008 Risico’s van medische technologie onderschat, 2008 [4] NVZ, NFU, Revalidatie Nederland, Convenant: Veilige toepassing van medische technologie in het ziekenhuis, 2011. [5] NVZ, NFU, Revalidatie Nederland, ZKN, Convenant: Veilige toepassing van medische technologie in de medisch specialistische zorg, 2016
Sanaz Berahmani, Rami Al-Dirini, David Wolfson, Dennis Janssen, Nico Verdonschot, Mark Taylor
Abstract: Bone formation at the bone-implant interface of an uncemented femoral knee implant relies on achieving adequate primary stability, which is commonly defined in terms of micromotions between bone and implant. Different patient, surgical and design variations can affect the micromotions. One important surgical variation is implant rotational alignment in axial (internal-external (IE) rotation) and frontal (varus-valgus (VV)) planes, which can be substantially varied due to the anatomic variability of bony landmarks and difficulty to identify them intraoperatively. The goal of this study is therefore to evaluate the effect of rotational alignments on the micromotions at the bone-implant interface of femoral knee implants. To this purpose, a database of 3D finite element (FE) models consists of 46 whole human femora incorporating patient-specific bone material properties was used. An algorithm was developed to select the correct implant size, resect distal femora and position uncemented Attune® femoral knee implant (DePuy Synthes, USA) on distal femora fully automated based on different pre-defined bony landmarks. Implant was rotated to attain five rotational alignments per bone model: default alignment based on manufacturer’s instruction (IE-3°/VV-5°), IE-0°/VV-0°, IE-0°/VV-7°, IE-7°/VV-0°, and IE-7°/VV-7°. Bone was fully constrained proximally and implant in all groups was subjected to the same implant-specific peak tibiofemoral load of gait cycle belonged to the default alignment, which proportioned based on the subject’s weight [1]. Micromotions at the bone-implant interface was predicted while a frictional (µ= 0.95) touching contact algorithm was assigned to the FE model. 95th percentile of micromotion distribution was used to evaluate the difference between groups. The averages of 95th percentile micromotions of all samples per group were 49± 19µm, 59±21µm, 41± 18 µm, 65± 25 µm, and 36± 17 µm (mean± SD) for the default, IE-0°/VV-0°, IE-0°/VV-7°, IE-7°/VV-0°, and IE-7°/VV-7°, respectively. Both groups with lower external rotations (IE-0°) had significantly higher micromotions and larger external rotation resulted in a significantly lower micromotions. Preliminary results showed that there is a negative correlation between micromotion and bone quality. The current study indicates the important effect of alignment on the primary stability. Our preliminary analysis has shown that the contact area, regional bone density, and patient’s weight are potential factors to explain current findings. The primary significance of the current study is that the developed technique can be used to evaluate implants in a larger scale and secondly it can be used as a guidance to assist orthopedic surgeon to decide upon the best alignment. [1] Fitzpatrick CK, Baldwin MA, Clary CW, et al. 2012. Evaluating knee replacement mechanics during ADL with PID-controlled dynamic finite element analysis. Comput Methods Biomech Biomed Engin 17:360–369.
Riza Bayoglu, Leo Geeraedts, Karlijn Groenen, Nico Verdonschot, Bart Koopman, Jasper Homminga
Abstract: Musculo-skeletal models enable clinical insights into the normal and pathological functioning of the spine. For example, the effect of thoracic kyphosis on spinal loads and trunk muscle forces can be studied [1]. Generic spine models are usually built by assuming anatomical symmetry between the left and right side of the human body. Therefore, it is critical to verify this assumption when building a model based on an anatomical dataset which was measured from only one side of the human body. In this study, we assess the validity of making a left-right symmetrical model by comparing the architectural parameters for the left and right sternocleidomastoid muscle. We measured the left and right sternocleidomastoid muscle (SCM) from an embalmed body of a 79 year-old male (height: 154 cm, mass: 51 kg). Prior to dissection, we divided both muscles into four muscle-tendon elements to represent their function more accurately. We dissected the muscles and stored them in 2% formaldehyde solution. Subsequently, we measured architectural muscle parameters: fiber length, tendon length, sarcomere length, optimum fiber length, pennation angle and physiological cross sectional area (PCSA), using the protocols described by Breteler et al. [2]. From each element, we used six to ten samples (single muscle fibers) for sarcomere length measurements with a He-Ne laser using the method of Cross et al. [3]. Although muscle fiber lengths were similar between the left and right, optimum fiber lengths were higher in the left muscle due to lower sarcomere lengths on this side. The total tendon length of the sternal part was also higher in the left SCM. The total PCSAs were 32% and 25% larger on the right side for the sternal and clavicular parts, respectively. These differences imply muscular imbalance for sternocleidomastoid muscle. The discrepancies in PCSAs will cause differences of 47 and 29 Newton in muscle force potentials (assuming a specific muscle tension of 90 N/cm2). Erroneous force predictions will develop in models if the presented differences are neglected. Although we measured one muscle in this study, similar-anatomical variations are anticipated for other muscles of the spine. REFERENCES [1] A.M. Briggs et al., “Thoracic kyphosis affects spinal loads and trunk muscle force”, Phys Ther, Vol. 87, pp. 595–607, (2007). [2] M. Breteler et al., “Measuring muscle and joint geometry parameters of a shoulder for modeling purposes”, J Biomech, Vol. 32, pp. 1191–1197, (1999). [3] H. Cross et al., “Comparison of methods for measuring sarcomere length in beef semitendinosus muscle”, Meat Sci, Vol. 5, pp. 261– 266, (1981).
L.W.A. Hermans, M.M. van Gilst, T. Weysen, T.R.M. Leufkens, Jan Bergmans, S. Overeem
Abstract: Chronic insomnia is a widespread problem, affecting about ten percent of the adult population1 . Insomniacs form a heterogeneous group. As a consequence, a substantial part of the patients does not respond to the generic cognitive behavioral therapy (CBT-I). Large discrepancies between objective macrostructural characteristics of sleep evaluated by polysomnography and subjective reports of complaints have been found in insomnia patients2 . Therefore, increasing emphasis has been given to study microstructural characteristics of the EEG. Current research on EEG microstructure indicates that patients with insomnia have decreased sleep stability, possibly reflecting hyperarousal2 . Here, we aim to better understand underlying mechanisms of insomnia and heterogeneity in treatment response, assessing macro- and microstructural EEG parameters. Eventually, this could lead towards finding clinically relevant subtypes, personalizing treatment and even predicting individual treatment success. Overall, 250 insomnia patients who are referred for CBT-I treatment to the Sleep Medicine Center Kempenhaeghe will take part in this study. Before the start of CBT-1, one night of ambulatory polysomnography (PSG) will be recorded at home. Cluster analysis will be used to detect groups of co-occurring macro- and microstructural EEG characteristics. We will focus on spectral power analysis, structural characteristics of REM-sleep, sleep protection mechanisms and (micro)arousals. We will evaluate these characteristics together, in order to study their relevance for phenotyping and their mutual relationships. Nowadays, spectral power and sleep stage characteristics can be extracted fully automatically. However, the detection of other microstructural parameters currently has to be done visually. Therefore, in this study automatic detection methods will be developed. For example, for sleep protection mechanisms we will focus the automatic detection of k-complexes, starting with simple methods based on amplitude thresholds. This can be extended using more complex methods, for example shape recognition. REFERENCES 1. Morin CM, LeBlanc M, Daley M, Gregoire JP, Merette C. Epidemiology of insomnia: Prevalence, self-help treatments, consultations, and determinants of help-seeking behaviors. Sleep Med. 2006;7(2):123-130. doi:10.1016/j.sleep.2005.08.008. 2. Riemann D, Spiegelhalder K, Feige B, et al. The hyperarousal model of insomnia: A review of the concept and its evidence. Sleep Med Rev. 2010;14(1):19-31. doi:10.1016/j.smrv.2009.04.002. 3. Parrino L, Ferri R, Bruni O, Terzano MG. Cyclic alternating pattern (CAP): The marker of sleep instability. Sleep Med Rev. 2012;16(1):27-45. doi:10.1016/j.smrv.2011.02.003.
Paul Henselmans, Paul Breedveld
Abstract: Natural Orifice Transluminal Surgery (NOTES) strives to reduce the invasiveness of surgery by using the body’s natural orifices (e.g. nose, mouth or anus) as the surgical entry point. Endonasal Skull Base Surgery (ESBS) is a NOTES procedure wherein the nose is used as an entry point for reaching lesions at the human skull base [1]. The skull base is a highly dense and complex anatomical area that cradles the brain and houses a network of delicate arteries and nerves. In such a dense environment, there is only limited space to maneuver the instruments. Currently, rigid instruments are used with a maximum diameter of 5 mm and preferably 3 mm. Because these instruments are rigid, they can only be inserted along a straight path from nose to lesion. A straight path through the dense skull base is not always possible, with the result that some lesions cannot be reached. This implies the need for instruments with a flexible shaft that can be inserted along a curved path. Such behavior is referred to as Follow the Leader (FTL) deployment, which can be compared with the locomotion of a snake [2]. The snake steers its head (the leader), whilst its body follows the same path. To enable FTL-deployment, the path has to be memorized. Typically, the path is memorized on a computer, and every degree of freedom of the flexible shaft is controlled by an individual actuator [3]. These actuators are generally placed inside the shaft, impeding miniaturization. A mechanical structure can also be used to memorize the path. This is for example done in an instrument called the HARP [4]. The main advantage of using a mechanical memory instead of a computer is that it reduces the number of actuators considerably. The HARP is, however, not suitable of ESBS due to its diameter of 12 mm. Further miniaturization of the HARP-design is challenging for the reason that its mechanical memory is placed inside the shaft. Therefore, we propose a novel mechanical memory mechanism for FTL-deployment that is situated outside the shaft of the instrument. REFERENCES [1] Verillaud, B., et al., Endoscopic endonasal skull base surgery. European Annals of Otorhinolaryngology, Head and Neck Diseases, 2012. 129(4): p. 190-196. [2] Choset, H. and W. Henning, A Follow-the-Leader Approach to Serpentine Robot Motion Planning. Journal of Aerospace Engineering, 1999. 12(2): p. 65-73. [3] Tappe, S., et al. Towards a follow-the-leader control for a binary actuated hyper-redundant manipulator. in IEEE International Conference on Intelligent Robots and Systems. 2015. [4] Ota, T., et al., A Highly Articulated Robotic Surgical System for Minimally Invasive Surgery. The Annals of Thoracic Surgery, 2009. 87(4): p. 1253-1256.
Niek Beckers, Demian Roelofs, Edwin van Asseldonk, Herman van der Kooij
Abstract: Assistance from a parent can help a child to learn how to ride a bicycle. Only until recently studies have been performed on whether physical interaction between humans can improve motor performance. One study in particular reported that being physically coupled to another person through a compliant connection improves motor performance of a simple movement task, regardless of motor performance level difference between interacting partners [1]. In real life however, the connection between two interacting persons is rigid, for instance when holding the saddle of a child’s bike or when shaking hands. In this study, we investigate whether rigidly coupled partners improve in a movement task better than one partner performing the task alone. Specifically, we tested whether physical interaction between a teacher – who is an expert at the task – and student improves the motor performance of the student faster compared to a student practicing the task alone. One teacher and 22 students were recruited and assigned to a dual group (with intermittent interaction with the teacher) and solo group (without interaction). Contrary to previous studies and expectations, the students who received intermittent physical interaction from the teacher did not show an increased motor performance over the students practicing the task alone. The performance of the teacher and student while interacting was significantly higher than when the student performed the task alone. Future work includes a detailed investigation on the effect of the stiffness of the physical connection, from rigid to compliant, on motor performance.
Ali Akyldiz, Lambert Speelman, Bas van Velzen, Jenny Dankelman, Frank Gijsen
Abstract: Atherosclerotic plaque rupture happens when plaque stresses exceed the tissue strength. Plaque stresses can be computed with finite element (FE) models [1], whose fidelity depends on the accurate representation of mechanical behavior of plaque tissue. The state of the art FE plaque models assume homogenous intima despite its highly heterogeneous structural composition. In this study, we evaluate the influence of intima heterogeneity on plaque stresses. FE models of 12 human coronary plaques were generated from histology images. Heterogeneous intima behavior was obtained by subdividing the intima in four clusters, based on histology image intensity (Fig. 1). A Young’s modulus between 6 and 891 kPa [2] was assigned to each cluster. Since the real stiffness of each cluster was unknown, 100 arbitrary stiffness combinations were simulated and the peak cap stress results were compared to the homogenous intima case (Young’s modulus=445 kPa). The relative standard deviation (SD) of peak cap stress of the heterogeneous models ranged from 14% to 75%, with an average of 32%. Compared to the homogeneous models, the lowest and highest peak cap stress values for the heterogeneous models were on average (±SD) -60% (±38%) lower and +87% (±48%) higher, respectively. This is the first study that incorporated intima heterogeneity in a computational plaque model. The results indicate a significant effect of intima stiffness heterogeneity on plaque stresses. The variation in peak cap stresses in heterogeneous models was not correlated to the median peak cap stress, so stress variation cannot be predicted from the cap stress level. Hence, the dataset provides valuable insight in the uncertainty and variation of the computed plaque stresses. REFERENCES [1] Akyildiz et al., Biomed Eng Online, 10:25, 2011 [2] Chai et al., J Biomech., 46(10):1759-66, 2013
Frédérique Meeuwsen, Jan Klein, Maarten van der Elst, Jenny Dankelman, John van den Dobbelsteen
Abstract: Electrosurgical devices are ubiquitous in modern operating and procedure rooms. Nowadays over 80% of procedures involve surgical energy. Its technique allows surgeons to skilfully dissect tissue and achieve hemostasis. However, there is a general lack of knowledge about the background theory and application methods among clinicians [1]. This makes the technology potentially unsafe and can cause fatal complications. Training in the use of this device is necessary to reduce complications and to enhance patient safety. In the past years this has been of great interest on a national level. Reports state that hospitals do not put enough effort into assuring the safe use of medical technology [2]. Our aim is to investigate how the training curriculum is currently organized and whether it needs improvement. All Dutch residents from six surgical subspecialties (general surgery, gynaecology, urology, thoracic surgery, plastic surgery and orthopaedic surgery) were invited to fill out a digital anonymous questionnaire about electrosurgery during August - December 2015. A total of 197 respondents completely filled out the questionnaire. Of them 69% had received training, mostly a single obligated theoretical lecture. When asking about their feeling of competency in the theory only 39% of residents was satisfied. Their feeling of competency in practical skills was rated higher with 71%. Moreover, 35% thinks the theoretical knowledge of their supervisors is not sufficient and 65% adjusts their way of application to the supervisor of the day. An interesting finding in this survey was the 30% that had encountered an incident (e.g. burn wounds) regarding electrosurgery. In conclusion, surgical residents often only get a single theoretical lecture about electrosurgery. They are not highly satisfied with their acquired competences in theory and practical skills regarding the device. Moreover, they are not satisfied with the theoretical knowledge of their supervisors. Since complications regarding surgical energy frequently occur, better and more frequent training for both residents and surgeons is needed to ensure patient safety. REFERENCES [1] Feldman LS, Fuchshuber P, Jones DB, Mischna J, Schwaitzberg SD, Force FT (2012) Surgeons don't know what they don't know about the safe use of energy in surgery. Surgical endoscopy 26:2735-2739 [2] Dutch Healthcare Inspectorate (IGZ) (2014) Veilig gebruik van medische technologie krijgt onvoldoende bestuurlijke aandacht in de ziekenhuizen.
Linda Eerikäinen, Lukas Dekker, Rik Vullings, Alberto Bonomi, Ronald Aarts
Abstract: Background: Post-extrasystolic potentiation (PESP) is an increase in the contractility of the heart following an extrasystolic beat [1]. Recently, the findings of augmented PESP in relation to heart failure have been reviewed [2]. Although further understanding of the relationship is needed, measuring PESP provides potential means to distinguish between failing and non-failing hearts. We propose a non-invasive method for continuous monitoring to evaluate the presence of PESP after a spontaneous extrasystolic beat using a photoplethysmographic (PPG) measurement from the wrist. Methods: Simultaneous electrocardiogram (ECG) and photoplethysmogram recordings from 7 patients that experienced at least 4 premature ventricular contractions (PVCs) during the night were analysed for the PESP detection. The ECG was recorded with a 12-lead Holter and the PPG and acceleration data with a wrist worn device. For the PESP analysis, the sequences of beat times including 8 beats preceding the extrasystole, the PVC, and the consecutive beat to the PVC were extracted from the ECG. The sequences that included artefacts, irregularities in the preceding beats or in the beats following the PVC were excluded from the analysis. The steepness of the rising slope of the PPG pulse was used as an indicator of force of the contraction. PESP was evaluated as the ratio between the steepness of the slope of the pulse after the PVC and the mean value for the steepness of the slope of the preceding pulses. Results: For one patient all the premature beat sequences were excluded from the analysis due to artefacts. In four patients there was an increase in the steepness of the slope for the majority of pulses following the PVC compared to the preceding pulses. This could be an indicator of the presence of PESP. In the remaining two patients a similar effect was not visible. Conclusions: Monitoring with photoplethysmography provides a promising method for a non-invasive continuous detection of post-extrasystolic potentiation. Further studies are needed to confirm our findings. REFERENCES [1] Langendorff O. Untersuchungen am Überlebenden Säugethierherzen. III. Abhandlung, Vorubergehende Unregelmässigkeiten des Herzschlages und ihre Ausgleichung. Pfluger Arch Physiol. Vol. 70, pp. 473-486, (1898). [2] Sprenkeler D. and Vos M., Post-extrasystolic Potentiation: Link between Ca2+ Homeostasis and Heart Failure? Arrhythmia & Electrophysiology Review. Vol. 5, pp. 20-26, (2016)
Joerik de Ruijter, Richard Lopata, Marc van Sambeek, Frans van de Vosse
Abstract: Rupture of carotid plaques accounts for 15 to 20% of ischemic strokes. To prevent a stroke patients undergo surgery, i.e. a carotid endarterectomy. However, only one out of six patients benefits from this intervention, so there is a significant overtreatment of patients. To develop better treatment criteria to improve patient diagnosis and clinical decision making, patient specific information of the plaque’s composition and mechanical properties is needed. In this study, a method was developed to estimate global material properties of the vascular wall using ultrasound imaging. Here, a finite element model (FEM) is matched with the displacement field measured by US, by updating the material properties of the FEM. As a result, a measure for arterial wall stiffness is obtained non-invasively. This inverse method was tested on artificial carotid phantoms made out of polyvinyl alcohol, with increasing complexity, both in geometry and the number of constituents. An experimental set-up was used to perform an inflation experiment and US imaging. This experiment mimics the imaging of a carotid artery in vivo. The inflation experiment is then simulated using finite element analysis: 3-D finite element meshes were constructed from 2-D ultrasound images. To estimate the displacements of the wall a 2-D iterative coarse-to-fine method was used. The error between the measured and simulated displacements was minimized using a downhill simplex optimization method. To validate the inverse approach, the mechanical properties of the PVA were determined using uni-axial tensile tests. For single layer phantoms, the estimated material properties (GFEM = 30 to 33 kPa) are within the range of the material properties found with the uni-axial tensile tests (GTT = 29 to 36 kPa). For stenosed double layer phantoms, we were able to estimate the stiffness of the outer layer and distinguish between a weaker inner layer and a stiffer outer layer. However, the absolute value of the estimated shear modulus of the inner layer was overestimated by 70% compared to the tensile tests. This can be explained by a construction error in the double layered phantoms and an underestimation of the displacements in the inner wall. In the present study, the material properties of a complete part of the material are estimated rather than a Young’s modulus distribution. However, in future work, this methodology will be extended to an inverse method to possibly identify the presence of lipid in the arterial wall, and monitor the development of plaques over time.
C.C. Roossien, J. Stegenga, Bart Verkerke, M.F. Reneman
Abstract: Office workers work for prolonged periods of time in a sitting posture, which may cause health problems [1]. A smart office chair with build-in pressure sensors can monitor sitting behavior and provide tactile feedback [2] [3]. Whether this improves sitting behavior and decrease discomfort is unknown. Aims of this study were to (1) explore the sedentary behavior of office workers, in and outside the smart chair; (2) investigate the effect of feedback on the sitting behavior and the effect of deactivation of the feedback signal; (3) explore the effect (number of) of the feedback signals on the sitting behavior; (4) investigate the effect of the feedback signal on the experienced local musculoskeletal discomfort (LMD) related to prolonged sedentary work. Participants were 45 office workers (26 females, 19 males, age 43.1±11.0 years) from 5 companies. This 12-week prospective cohort study was split into four phases (ABCB-design); (1) acclimation, (2) monitoring I, (3) intervention, (4) monitoring II. The sitting behavior on the chair was monitored continuously using the BMA Axia Smart Chair. Once per phase the subject wore an activity tracker, Actigraph, and received a questionnaire about experienced musculoskeletal discomfort. The first results show that these office workers spend on average 7.1 h per working day in sitting position of which 5.4 h at their own working place. The subjects sat on average 12.9% of their working day in an optimal supported position. During the monitoring phase II, sitting in an optimal supported position lowered significantly, as did the LMD. This first results of this study shows that the sedentary behavior and LMD did not improve during the intervention phase. Although this smart office chair is a good non-obstructive method to monitor sedentary behavior in the chair, the feedback signal needs to be optimized. REFERENCES [1] Thorp, A.A., et al. „Prolonged sedentary time and physical activity in workplace and non-work contexts: a cross-sectional study of office, customer service and call centre employees,” International Journal of Behavioral Nutrition and Physical Activity, vol. 9, nr. 128, pp. 1-9, 2012. [2] Goossens, R.H.M., et al. „An office chair to influence the sitting behavior of office workers,” Work: A Journal of Prevention, Assessment and Rehabilitation, vol. 41, nr. 1, pp. 2086-2088, 2012.
Gokhan Gunay, Mann Ha Luu, Adriaan Moelker, Theo van Walsum, Stefan Klein
Abstract: Percutaneous radiofrequency ablation is a method for liver tumor treatment when traditional surgery is not an option [1]. It is a minimally invasive treatment and may be performed under CT image guidance when the target tumor does not give sufficient contrast on ultrasound images. Registration of the pre-operative contrast enhanced CT image to the intra-operative CT image is hypothesized to improve guidance. This is a highly challenging registration task due to differences in patient poses and thus large deformations. In this study, we introduce a semi-automated registration algorithm to be used when fully automated registration algorithms give unsatisfactory results. The method is based on a conventional nonrigid intensity-based registration framework [2], extended with a novel point-to-surface penalty. The point-to-surface penalty serves to improve the alignment of the liver boundary, while requiring minimal user interaction during the intervention: annotating some points on the liver surface at those regions where the conventional registration seems inaccurate. The method is evaluated on 18 clinical datasets and shown to improve registration accuracy compared with the conventional nonrigid registrations.
Ronaldo Martins da Ponte, Vasiliki Giagka, Wouter Serdijn
Abstract: Implants for neural recording or stimulation require arrays of microelectrodes to sense the signals or steer the stimulus current to the targeted nerve fibers or to specific brain circuits. In some applications, a higher number of microelectrodes is crucial to achieve the required high spatial-temporal resolution [1]. However, the larger the number of electrodes on the interface, the larger the number of long routing tracks to connect the external electronics, increasing the implant size and, often, compromising its reliability. To overcome this problem, we propose an implant containing electronics distributed over the microelectrode array for data routing, so the number of routing tracks can be drastically reduced. However, unlike previous works [2], the active electrodes herein proposed are capable of exchanging data over the array, such as mode selection of electrodes, communicating via a distributed network. The implant is also capable of communicating wirelessly with the outside world, receiving control information on magnitude and timing for the stimuli currents and providing feedback on the neuronal electrical activity, before, during and after a stimulation event. To cope with the power demand and chronic implantation, the fabrication of an antenna on a flexible substrate, combined with a power-management unit (PMU) and an energy storage element, is proposed, providing energy transfer [3] to the distributed electronic devices over the microelectrode array. In addition, the proposed implant will be fabricated using flexible materials since they have mechanical properties closer to those of living tissue, thus reducing the risk of inflammation and tissue damage.
Guangfei Zhang, Ihor Kirenko, Caifeng Shan, Ronald Aarts
Abstract: Pulse Transit Time (PTT), the time delay for the blood pressure wave to travel between two arterial sites, can be used to estimate Blood Pressure (BP) [1]. Blood pressure can be measured by an invasive way, e.g a catheter, on a cuff. For an unobtrusive way, photoplethysmography (PPG) was utilized as a potential alternative since it records the waveform of blood volume variation which is correlated with arterial pressure waveform [1]. Currently, contact PPG (cPPG) and imaging PPG (iPPG) are two feasible ways to obtain the PPG signal [2]. A collaborative system using both of them enables flexibility into unobtrusive Blood Pressure measurement since such system can be housed on a smart watch by taking advantage of the existing PPG sensor, facilitating the measurement of BP in an inconspicuous way. In this study, we investigated that new blood pressure measurement system (ICPPG system) comprised of cPPG and iPPG sensors. Experiments were carried out on 30 subjects (21 males and 9 females) for collecting their iPPG waveform from the face and cPPG waveform at fingertip synchronously during several sessions with and without physical exercises. Blood Pressure for each session was measured by a commercial BP meter with an electronic inflatable cuff placed on the left upper arm of the subject. Simultaneously, as a reference, an extra pair of cPPG signals from temple and wrist was also collected by a commercial physiologic monitor. Among 30 subjects, our ICPPG system is able to produce reliable PTT values which are consistent with the range of PTT from previous results [3]. Moreover, it was verified that the PTT was strong correlated (r>0.6) with Systolic Blood Pressure (SBP) for 77% subjects, Mean Blood Pressure (mBP) for 67% subjects and Diastolic Blood Pressure (DBP) for 57% subjects. In conclusion, a reliable PTT can be produced by our new system. Using the PTT to estimate Blood Pressure in a unobtrusive way is feasible. REFERENCES [1] Ramakrishna Mukkamala, Jin-Oh Hahn, Omer T. Inan et al., “Toward Ubiquitous Blood Pressure Monitoring via Pulse Transit Time: Theory and Practice,” IEEE Trans. Bio. Eng., vol. 62, no. 8, pp. 1879-1901, Aug. 2015. [2] Yu Sun, Nitish Thakor, “Photoplethysmography Revisited: From Contact to Noncontact, From Point to Imaging,” IEEE Trans. Bio. Eng., vol. 63, no. 3, Mar. 2016 [3] In Cheol Jeong, Joseph Finkelstein, “Introducing Contactless Blood Pressure Assessment Using a High Speed Video Camera”, Springer Science & Business Media New York 2016, pp. 40-77, Jan. 2016
Yageng Li, M.A. Leeflang, Behdad Pouran, Jie Zhou, Holger Jahr, Amir Zadpoor
Abstract: For the treatment of large bone defects, no optimal solution suitable for clinical applications has yet been found. Porous metal orthopaedic implants have been regarded as a ground-breaking invention in skeletal reconstructive surgery, thanks to their ability to allow bone tissue regeneration inside the scaffold structure. Porous titanium implants with good biocompatibility and mechanical properties have been extensively investigated. However, such implants remain permanently in the body, which may cause physical irritations and chronic inflammatory local reactions[1]. Biodegradable metal scaffolds are expected to achieve the same mechanical and biological functions as permanent implants, but to disappear completely when they are no longer needed[2]. This project will be focused on investigating the biodegradation behaviour and mechanical properties of additively manufactured magnesium scaffolds. The magnesium alloy WE43 containing 3.7-4.3% yttrium, 2.4-4.4% rare earth elements and 0.4% zirconium was the first alloy subjected to selective laser melting (SLM) trials to fabricate scaffolds. Analyses of microstructure, surface morphology and chemical composition were performed using Scanning Electron Microscopy/Energy Dispersive Spectroscopy. Phase identification by X-ray Diffraction was performed on the cross section of SLM scaffolds. Static mechanical testing of samples was carried out in accordance with the ISO 13314:2011 standard. WE43 magnesium alloy scaffolds with 70% porosity could be produced by means of SLM. They had a fine microstructure with oxide particles embedded. The average grain size was around 5 μm. Young’s modulus of the as-built magnesium scaffolds was 2 GPa and compressive yield strength 40 MPa, which are comparable to those of cancellous bone. In the follow-up research, immersion tests and electrochemical experiments will be performed to investigate their in vitro degradation behavior. Static compression and fatigue test will also be conducted at different corrosion time points. REFERENCES 1. Moravej, M. and D. Mantovani, Biodegradable Metals for Cardiovascular Stent Application: Interests and New Opportunities. International Journal of Molecular Sciences, 2011. 12(7): p. 4250. 2. Chen, Y., et al., Recent advances on the development of magnesium alloys for biodegradable implants. Acta Biomaterialia, 2014. 10(11): p. 4561-4573.
Nazlı Tümer, Frans Vos, Marjon Stijntjes, Geert Streekstra, Christian Askeland, Gabrielle Tuijthof, Amir Zadpoor
Abstract: ABSTRACT Introduction: (Osteo)chondral defects (OCDs) can lead to chronic joint pain, decreased level of activity, and osteoarthritis due to poor self-healing ability [1]. Long-term monitoring of patients with OCDs can be helpful to understand the nature of cartilage regeneration in vivo and formulate effective treatment strategies. In follow-up studies, use of ultrasound (US) imaging modality is preferable as it is non-invasive and cost-effective. Nevertheless, US imaging has several limitations that can affect evaluation of OCDs. To improve post-operative assessment of defects (i.e. evaluation of changes in OCDs due to treatment), US images of the patient can be fused with a pre-operative CT scan of the same patient. In this study, we propose a new rigid surface-volume registration scheme to accurately register CT and US images of the ankle joints, which is the second most common site for OCDs. Methods: Points on the ankle bone surfaces are first extracted from CT data. Subsequently, local phase features are calculated from the monogenic signal representation of 2D US images of the ankle to highlight the bones in the US data. 3D monogenic signal data is reconstructed from the 2D data using the position of the transducer that is recorded with an optical tracking system. The registration of the surface from the CT image to the monogenic signal feature volume is initialized at a location that is not too far from the optimum based on six fiducial markers applied during image acquisition. The objective function to be optimized during registration is the sum of monogenic signal features over the transformed surface extracted from the CT data. Six transformation parameters (three for rotation and three for translation) are estimated using the covariance matrix adaptation evolution strategy [2]. Results: The registration scheme was tested on CT and US images of 14 cadaveric ankle joints acquired at the Academic Medical Center Amsterdam. It was visually observed that CT and 3D US images can be successfully registered. Conclusion: Preliminary results show that the proposed registration scheme works well. Further investigation and quantification is required to evaluate the accuracy and precision of the registration. REFERENCES [1] N. Tümer, L. Blankevoort, M. Van De Giessen, M. P. Terra, P. A. De Jong, H. Weinans, G. J. M. Tuijthof, and A. A. Zadpoor, “Bone shape difference between control and osteochondral defect groups of the ankle joint,” Osteoarthr. Cartil., pp. 1–8, 2016. [2] N. Hansen and A. Ostermeier, “Completely Derandomized Self-Adaptation in Evolution Strategies,” Evol. Comput., vol. 9, no. 2, pp. 159–195, 2001.
Rob Zink, Alexander Bertrand, Sabine Van Huffel, Maarten De Vos
Abstract: In the past few years, electroencephalography (EEG) has been employed for auditory attention detection (AAD). AAD holds promising potential for usage in auditory-assistive devices. A pretrained decoder that makes a linear combination of the EEG and delayed versions of it allows to extract a signal with a significantly higher correlation to an attended speaker’s envelope as compared to an unattended speaker’s envelope [1]. In such dual-speaker scenarios, the subject’s attention to a specific speaker could be reliably detected in research labs [1, 2, 3, 4]. Being able to train subjects in achieving high(er) AAD performance would increase the application potential of AAD. This requires an acceptable temporal resolution and the analysis should take place online. In the current work, we explored the application of the AAD in near real time. The subjects were recorded in an office environment with mobile EEG hardware and consumergrade headphones. This is more convenient for the subjects compared to a lab environment. In addition, we apply a neurofeedback scenario to provide a proof-of concept for a fully automated closed-loop system. We achieved high AAD accuracies with a trial length of 10 seconds and provided subjects with visual feedback on their ongoing performance. The accuracies in the present work were obtained with 24 ‘wet’ electrodes and were found to be stable up to removal of 16 channels. These results are in line with insights presented by [4] and are encouraging for future work in online AAD processing. This exploratory study proves the feasibility of investigating the effect of neurofeedback in such a setting, paving the way for future studies (e.g. long-term training at subjects’ home).
Mustafa Radha, Ronald Aarts
Abstract: Systolic blood pressure (SBP), which is the arterial pressure during right after ventricular ejection, is a widely used cardiovascular measure. However, existing methods to measure SBP, continuously are highly invasive or obtrusive. Therefor pulse wave velocity (PWV), the speed of the arterial pulse wave, has been proposed as a surrogate measure for SBP. PWV is derived by measuring the pulse arrival time, usually between electrocardiogram (ECG) and photoplethysmogram (PPG). However, the relationship between PWV and SBP is not the same in all subjects: there is a person-specific linear calibration required [1]. This calibration is attributed to specific arterial properties such as the elasticity, vessel thickness and radius, which all contribute to systemic vascular resistance (SVR). Since SVR cannot be directly measured, no clear solution exists for the calibration other than a high number of simultaneous measurements of blood pressure and pulse wave velocity through which linear regression coefficients a and b are derived of the form BP = a *PWV + b. Here, a would correspond to SVR. We propose to measure SVR by examining the response of blood pressure to a physical stressor, in particular physical exercise. Physical stressors increase the demand for blood in the engaged muscles, which the heart responds to with a higher cardiac output (CO). According to the two-element Wind-Kessel model of circulation, an increase in CO will also increase BP, relative to the resistance R: BP = CO * SVR. Thus, when a certain physical exercise is performed, the increase in BP will be proportionate to SVR, which can be used to calibrate the relationship between SBP and PWV. Sixteen subjects were enrolled to cycle at a fixed moderate pace on a stationary ergometer. Continuous SBP through vascular unloading was measured, as well as PWV from the ECG to wrist PPG (in a bracelet). The parameters a and b were calculated for each subject through regression. The response of SBP was measured as the difference between two measurement points of SBP, one during rest and the other during cycling. We show that the SBP response to the physical exercise is highly correlated with a (Pearson’s correlation of 0.95). We conclude that the response to physical stressors is indicative of SVR and thus can be used to determine person-specific calibration parameters between SBP and PWV.
Kasper Marstal, Floris Berendsen, Marius Staring, Stefan Klein
Abstract: In the past decade there has been an increasing interest in relating information from medical imaging data, spurred by a growing availability of scanners, modalities and computing power. This typically involves image registration for transforming images into a common coordinate system such that corresponding pixels represent homologous biological points. Clinical applications include segmentation of anatomical structures, computer-aided diagnosis, monitoring of disease progression, surgical intervention and treatment planning. A significant amount of research has focused on developing the registration algorithms themselves, while little attention has been paid to accessibility, interoperability and extensibility of these algorithms. Scientific source code is typically not published, is lacking documentation or is difficult to use because it was not written for fellow researchers. This is a problem since image registration is a prerequisite for a wide range of image analysis tasks. Open source, user-friendly implementations of scientific software make state-of-the-art methods accessible to a wider audience, promote opportunities for scientific advancement, and support the fundamental scientific principle of reproducibility. To this end, we have developed the SimpleElastix software package. Elastix is a popular, open source, command-line program for intensity-based registration of medical images that allows the user to quickly configure, test, and compare different registration methods [1]. SimpleElastix is an extension of SimpleITK [2] that combines SimpleITK with Elastix, enabling the use of Elastix algorithms directly in convenient Python, Java, R, Octave, Ruby, Lua, Tcl and C# development environments on Linux, Mac and Windows. This allows the user to focus more on the registration problem and less on command line calls and the underlying C++ implementation. In addition, we have developed an extensive test suite, online documentation, and a citable online software repository following recommended best practices for scientific software publication, all within an online environment that stimulates community involvement and contribution. The user configures the registration procedure via a parameter object containing key-value pairs that atomically define the registration method, its components, and any settings it might require. This allows users without expert programming knowledge to compare different registration algorithms and customize registration procedure to their own data. As means of demonstration, we apply the method to two registration problems and show how to setup experiments using a minimal amount of code. The software is open source, licensed under the permissive Apache 2.0 license, and available at https://github.com/kaspermarstal/SimpleElastix. [1] S. Klein, M. Staring, K. Murphy, M.A. Viergever, J.P.W. Pluim, “Elastix: a toolbox for intensity-based medical image registration”, IEEE Trans Med Imag, 29:196-205, 2010. [2] B. C. Lowekamp, D. T. Chen, L. Ibáñez and D. Blezek, The Design Of SimpleITK, Front Neuroinform, 30;7:45, 2013