• Home
• Preliminary Program
• Call for abstracts
• Conference topics
• Important dates
• Job market
• Keynote speakers
• Awards
• Sponsors & Exhibition
• Registration & Venue
• Organising board
• Contact
• Privacy policy

16:00   Poster session II Analyzing Freezing of Gait Using Foot Switch Data Jamie Jansen, Ying Wang, Mike Cohen, Jorik Nonnekes, Richard van Wezel Abstract: In Parkinson’s disease (PD) a poorly understood symptom is Freezing of Gait (FoG), which is defined as a brief, episodic absence or marked reduction of forward progression of the feet despite the intention to walk. This symptom affects many PD patients in daily-life by increasing anxiety, the risk of falling and the quality of life. Despite the fact that during a freezing episode the feet are felt to be ‘glued’ to the floor, it is not necessarily a frozen akinetic posture. Forces under the legs can change during FoG with temporal structure and organization (Hausdorff, Balash & Giladi, 2003). New insights and technological improvements might improve quality of life by prediction and detection of FoG. Current methods, such as cueing, are continuous and are thought to work even better when only activated on-demand. To explore this, an experiment was performed to acquire and implement EEG, ECG, motion sensor and foot switch data. Here, the goal was to provoke FoG by stepping movements in place in 15 PD patients (Hoehn and Yahr stage 2-4) with FoG during an “off” medication state. The movement-in-place tasks were composed of 3 conditions: stepping in place, a normal 180-degree turn and a rapid turn. In this particular part of the study, analysis of the foot switch data is performed. In the experiment, four foot switches were attached to four locations under both feet and activation was measured over time. Analysis of foot switch data and an algorithm that detects and FoG will be presented. Automated Multimodal Epileptic Seizure Detection Using EEG and ECG Kaat Vandecasteele, Thomas De Cooman, Ying Gu, Steven Vandeput, Evy Cleeren, Kasper Claes, Jonathan Dan, Wim Van Paesschen, Sabine Van Huffel, Borbála Hunyadi Abstract: Seizure diaries kept by the patients are subjective and often unreliable. In order to obtain an objective seizure diary outside the hospital, an automated wearable seizure detection device is required. To detect non-convulsive seizures, electroencephalography (EEG) and electrocardiography (ECG) are interesting biomedical signals. Instead of using the full scalp EEG, only behind-the-ear EEG channels are used here, which can be recorded with a wearable device outside the hospital. This research proposes a patient-specific automated seizure detection algorithm based on behind-the-ear EEG and HR. The proposed algorithm combined 4 behind-the-ear EEG channels (two behind each ear) with the heart rate. These signals were recorded with the hospital hardware. The algorithm consisted of two sequential steps. Firstly, a Support Vector Machine-based algorithm generated alarms based on the EEG data only. Secondly, on the data segments responsible for generating these alarms, a k-nearest neighbors-based algorithm, using EEG and ECG data, was run. A key element in this algorithm is the alignment in time of the ictal manifestation in EEG and ECG. This algorithm was evaluated on a dataset of consecutive patients recorded in the University Hospitals Leuven between with the following inclusion criteria (1) at least 2 seizures with EEG correlates (2) at least an average ictal HR increase of 30 beats per minute. The dataset consists of 1897 hours of data originating from 18 patients including 80 focal impaired awareness seizures, arising from (fronto-) temporal (78) and frontal-parietal (2) lobe. The average Sensitivity (Sens), False Positives per hour (FP/h) and Positive Predictive Value (PPV) were calculated for the EEG-based and the EEG/ECG-based seizure detection algorithm. The EEG-based algorithm resulted in a Sens of 78.36%, 1.09 FP/h and a PPV of 0.13, whereas the EEG/ECG-based algorithm resulted in a Sens of 65.73%, 0.15 FP/h and a PPV of 0.31. Videolaryngoscopy: Barriers, Design Requirements and Context-Specific Design for Low- and Middle-Income Countries Julie Fleischer, Linda Wauben, Jan Klein, Jenny Dankelman Abstract: Introduction: Difficult intubation is a well-known clinical problem that is most often encountered in emergency medical settings, but it also complicates airway management in intensive care units and operating theatres. Difficult intubations are associated with complications, such as trauma, hoarseness and hypoxia. Videolaryngoscopy can reduce intubation time, increase first-pass success rate and reduce the number of oesophageal intubations. However, videolaryngoscopes are not widely used in low- and middle-income countries (LMIC), increasing the risk of complications and possibly making difficult intubation an obstacle to perform surgery. Therefore, this study identifies barriers to videolaryngoscopy in LMIC as well as stakeholders’ needs and generates design requirements for a new context-specific design. Method: Twenty-seven questionnaires were returned by surgeons, anaesthesiologists, biomedical engineers, sterilisation department employees and other healthcare professionals attending the 2018 Surgical Society of Kenya Conference and 30 questionnaires were returned from three hospitals in Kenya. Current practice concerning use and maintenance of airway equipment was observed for two days in a Kenyan hospital and non-structured interviews were done with stakeholders from multiple institutions in Kenya. Moreover, a literature search was performed to compare performance of various videolaryngoscope designs. Based on these requirements a reusable videolaryngoscope was developed. Results: Barriers to widespread use of videolaryngoscopy in Kenya include high procurement costs, difficulty in obtaining disposables or proprietary spare parts and unfamiliarity with the equipment. The most important design requirements include easy manoeuvrability and a short learning curve so that it can be used by experienced and inexperienced providers in all settings, as well as the possibility for local repair. A design was created that accommodates both standard and hyper-angulated blades to manage various patient anatomies. This multi-blade videolaryngoscope combines a low-cost USB endoscope with a smartphone screen. Discussion & conclusion: The identified barriers and design comparison led to a prototype for a low-cost and easy to use videolaryngoscope. The next steps include a functional evaluation for direct and indirect laryngoscopy of the blades, an analysis of the light source and image quality and an assessment of maintenance and cleaning procedures in the local context, resulting in a device that could increase use of videolaryngoscopy in LMIC. The Feasibility of Hands-Free Acquisitions of Skeletal Muscles and the Heart During Exercise Marloes Sjoerdsma, Maarten Heres, Thijs Schoots, Frans van de Vosse, Richard Lopata Abstract: Continuous, dynamic ultrasonic evaluation of architectural and mechanical parameters of skeletal muscles is challenging. Additionally, intra-exercise measurements are generally more relevant compared to post-exercise parameters. Field-Of-View (FOV) consistency is often essential for measurement accuracy. However, probe orientation and position are challenging to reproduce and preserve. Probe fixation might improve FOV stability and enable continuous acquisitions during exercise. In this study, the feasibility of hands-free ultrasound was investigated in two distinctively challenging fields, i.e. myosonography and echocardiography. The FOV stability obtained by a fixated probe was compared to trained sonographers. In myosonography, longitudinal images of the vastus lateralis muscle were acquired in ten healthy volunteers whilst cycling. In echocardiography, eight exercise stress tests were performed. For both, a MyLab 70 Esaote system was employed. To overcome inter-observer variability and bias, feature-based algorithms were implemented to automatically detect and measure muscle diameter and pennation, and interventricular septum curvature and tilt. The FOV stability was assessed based on the temporal deviation in these function-related parameters and the structural similarity, measured using the Complex-Wavelet Structural Similarity Index Method (CW-SSIM). In myosonography, the CW-SSIM indices of the muscle tissue of the hands-free acquisitions were significantly superior compared to two trained sonographers (0.72 vs 0.59 and 0.67). The hands-free obtained muscle pennation and diameters were at least as consistent as the manual acquisitions. In echocardiography, the hands-free CW-SSIM indices were at least equal to the manual measurements (apical: 0.78 vs 0.66, and parasternal: 0.64 vs 0.63). However, the parasternal CW-SSIM variation of the observer was lower in comparison to the hands-free acquisitions, due to manual corrections made for respiration induced cardiac dislocation. In the future, these dislocations could be tracked and corrected automatically using a software steerable component implemented into the probe fixation device. The variation in septum curvature and tilt appeared smaller in the hands-free data (0.25 vs 0.61 mm-1 and 0.28 vs 0.58º, respectively). Hands-free, continuous, dynamic acquisitions promise to be beneficial in many clinical fields and sports. Moreover, it has the potential to decrease work-related musculoskeletal afflictions in sonographers. Atomic Scale Surface Engineering of Micro- to Nano- Sized Pharmaceutical Particles Damiano La Zara, Di Zhang, Mike Quayle, Gunilla Petersson, Staffan Folestad, Ruud van Ommen Abstract: Active pharmaceutical ingredients (APIs) and pharmaceutical excipients are typically small solid organic particles present in the form of powders. The morphology and surface characteristics of both APIs and excipients play a crucial role in both their manufacturing and administration into the human body. Presently, nearly 40% of APIs in the pharmaceutical market have poor water solubility (i.e., below 0.1 mg mL-1), which then results in limited dissolution rate and low bioavailability into the human body. One of the best approaches to improve the bioavailability of poorly water-soluble drugs is particle size reduction, including micronization which delivers particles down to the micrometer or, in some cases, nanometer size. However, micronized particles are highly cohesive and moreover might present amorphous regions that crystallize over time in an uncontrolled manner. Therefore, there is an unmet need for commercially viable solutions to provide surface modification of pharmaceutical powders that lead to improved processability and stabilization of solid state forms (e.g., amorphous, polymorphs, hydrates) or clinical benefits, e.g., controlled release of API from dosage forms. In this work, we demonstrate the use of atomic layer deposition (ALD) as a route to modify the properties of both API (i.e., budesonide) and excipient (i.e., lactose) particles, namely dissolution rate and dispersibility. In brief, ALD enables the growth of uniform and conformal ultrathin films on complex 3D structures with exquisite thickness control at the sub-nanometer level. We study the growth of Al2O3 nanofilms in terms of uniformity and conformality on both crystalline and amorphous pharmaceutical particles of different sizes. The deposition quality strongly depends on (i) the crystal structure of the particles, (ii) the fluidization behaviour of the particles, and (iii) the chemical reactants used in the coating process. Time-of-flight secondary ion mass spectrometry and transmission electron microscopy reveal the deposition of uniform and conformal nanofilms on large crystalline particles, whereas uniform but non-conformal nanofilms are observed on small partly-amorphous particles. In-vitro dissolution tests reveal more sustained release with increasing film thickness, and in particular with improved coating uniformity. Finally, the Al2O3-coated particles exhibit considerably higher dispersibility in liquid phase, which suggests higher bioavailability than the uncoated ones. The Development of Predictive Simulations to Assess the Effects of Impairments on Gait in Children with Cerebral Palsy Kirsten Veerkamp, Thomas Geijtenbeek, Jaap Harlaar, Marjolein van der Krogt Abstract: Children with cerebral palsy (CP) often show problems with gait due to both neural and non-neural impairments [1]. Optimal treatment selection is challenging because of the complex, multilevel nature of these problems. Neuro-musculoskeletal modelling might provide a valuable source of information to assist in treatment selection, by having potential to predict post-treatment outcomes. However, often generic models are used, which are based on healthy adults and therefore their use may not be accurate in a clinical population like CP. Hence, an essential step forward is to implement patient-specific features into the model. The aim of this project is to develop and validate predictive gait simulations incorporating various impairments as present in CP. Recently, a platform for forward dynamic, predictive gait simulations has been developed (SCONE https://simtk.org/projects/scone). This platform implements a two-dimensional OpenSim [2] model with realistic Hill-type muscles and includes various customizable control strategies, allowing to create a stable and realistic gait pattern. The kinematics (joint angles), kinetics (ground reaction forces, joint moments and joint powers) and muscle activation patterns predicted by the forward dynamic simulations of the model will be compared to experimental gait data. Various impairments that are common in CP (i.e. muscle weakness, spasticity and joint contracture) will be implemented into the model separately. Experimental gait data of people with ankle plantar flexor weakness, ankle spasticity and with increased ankle joint stiffness imposed by kinesiotape will be used for validation, respectively. Finally, all three of these impairments will be combined in the model and compared to gait data of patients showing varying levels of weakness, spasticity and contracture. Statistical parametric mapping will be used to compare the model predictions and experimental data. This project will provide fundamental insights into the effects of various common impairments in CP on gait performance. This knowledge can be valuable for understanding the underlying pathologies. In addition, the development of patient-specific predictive simulations may be an important step forwards to using neuro-musculoskeletal modelling as a predictive tool in clinical practice, with the ultimate goal of improving the challenging treatment selection in patients with complex neurological disorders. References [1] Graham et al., Nature Reviews Disease Primers (2016) [2] Delp et al., IEEE Transactions on Biomedical Engineering (2007) Evaluation of Inter- and Intra-Operator Reliability in Manual Segmentation of Femoral Metastatic Lesions Ali Ataei, Florieke Eggermont, Milan Baars, Yvette van der Linden, Nico Verdonschot, Esther Tanck Abstract: Bone is the third most commonly affected tissue by metastases. Metastases in the femur are often painful and increase the risk of pathological fracture. Accurate delineation of bone metastases is amongst others important for radiotherapy planning and determining exact radiation fields. Hence, the aim of this study was to investigate the inter- and intra-operator reliability in manual segmentation of femoral metastatic lesions. CT-scans of 54 patients with femoral metastases (19 osteolytic, 17 osteoblastic and 18 mixed) were included in this study. Two inexperienced operators were trained and then asked to segment metastatic lesions in all patients twice with a four-week time interval. Dice coefficients (DC) were calculated to quantify the inter- and intra-operator overlap of the segmentations. A DC˃0.8 indicates good reliability and a DC between 0.6 and 0.8 is acceptable. Mean DCs for inter-operator reliability were 0.54 (±0.28) and 0.50 (±0.32) for the first and second time segmentation, respectively. Mean DCs for intra-operator reliability were 0.56 (±0.28) for operator I and 0.71 (±0.23) for operator II. For large lesions (over 20000 voxels), the mean DC was 0.82 (±0.07), whereas for smaller lesions (less than 20000 voxels), the mean DC was 0.60 (±0.28). Osteoblastic lesions had the highest mean inter- and intra-operator DC with 0.57 in the first time, 0.60 in the second time, 0.60 for operator I and 0.78 for operator II compared to osteolytic (0.54, 0.45, 0.56, 0.63) and mixed lesions (0.51, 0.46, 0.52, 0.74). The mean DC of neither inter- nor intra-operator reliability test were satisfying. Only one of the operators showed an acceptable intra-operator reliability. Additionally, segmentation of larger lesions resulted in good inter- and intra-operator reliability. Higher reliability for osteoblastic lesions can be justified by their more clear edges on CT-scan images. This study reveals that the current manual segmentation approach might results in unacceptable reliability rates. Hence, development and use of an automated segmentation method might be beneficial. Mechanical Predictors of Rupture in Atherosclerotic Plaque: Beyond "where Stress, There Rupture" Daniel Rutten, Ronald van den Berg, Frank Gijsen, Ali Akyildiz Abstract: The leading cause of death worldwide is acute cardiovascular event precipitated by atherosclerotic plaque rupture1. Rupture occurs when the structural integrity of plaque tissue is compromised by an overwhelming mechanical load. Plaque biomechanics are therefore key to predicting rupture. Seminal studies demonstrated colocalisation of rupture loci and mechanical stress concentrations in plaque-models2,3. Subsequent research has mostly consisted of improving modelling techniques and the acquisition of (imaging-)data. However, there has been little-to-no investigation into the rupturepredictive power of alternative mechanical metrics. This study performs colocalisation-analyses for a comprehensive selection of mechanical metrics in order to identify appropriate metrics for rupture prediction and improve our fundamental understanding of rupture mechanisms. A total of ten stress, strain, and energy metrics were selected for isotropic and anisotropic FEMs based on a unique histopathological dataset of ruptured carotid plaques (n=30)4. The plaques had undergone minimal morphological change during rupture, allowing accurate replication of pre-rupture geometry. Segmentation of the histology images produced plaque geometries for isotropic FEMS. In addition, histology images were put through a custom-made image-processing tool capable of detecting intra-plaque collagen fibres. Local fibre orientations and dispersions were implemented using the Holzapfel-Gasser-Ogden constitutive model, creating a second set of histology-based, heterogeneous and anisotropic FEMs. Colocalisation analyses of rupture sites with selected metrics were performed in the isotropic and anisotropic FEMs. Ten of the twelve ruptures (83%) analysed showed colocalisation with at least one metric. Two metrics had sensitivity below 50% and the average sensitivity of the metrics was 58% (64% without the two <50% metrics). Although isotropic and anisotropic metrics had comparable performance overall, anisotropic fibre-shear strain had the highest sensitivity (75%). Furthermore, it was the only predictor for one rupture. The success of anisotropic fibre-shear strain supports the previously-posited delamination-based damage mode4. Furthermore, reliable strain-based predictors may promote the application of biomechanical results in clinical settings, since strain measurements are possible with recent advancements in ultrasound and MR imaging. Planned analysis of the remainder of the dataset is expected to strengthen these results. An Ultra High-Frequency 8-Channel Neurostimulator Circuit with 68% Peak Power Efficiency Alessandro Urso, Marijn Van Dongen, Wouter Serdijn Abstract: In order to recruit neurons in the targeted tissue, constant-current neural stimulators are usually used. Recently, Ultra High-Frequency (UHF) stimulation has been proposed and proved to have the same efficacy of constant current stimulation [1]. The total number of external components is reduced, while the power efficiency is increased. This leads to a smaller stimulator device with an increased battery life. The core circuit of the UHF neurostimulator is a DC-DC converter, which generates current pulses. Each stimulation phase is made of a burst of current pulses injected into the tissue at a determined frequency. The amplitude of the pulses is controlled by means of a duty cycle signal. Here, we present the design guidelines and the IC measurement results of a power-efficient UHF neural stimulator. An overall peak power efficiency of 68% is achieved when 8 independent channels with 16 fully configurable electrodes are used. The only external component is an inductor. It is operated in a time-interleaved fashion across all the activated channels. A novel zero-current detection scheme is proposed. It does not require the freewheel diode usually used in DC-DC converters to prevent current flow from the load back to the inductor. A gate-driver circuit is implemented. It allows to use thin gate oxide transistors as high voltage switches. By doing so, the external high voltage supply, usually used in neural stimulators, is avoided and the neurostimulator is powered from a 3.5 V input voltage. Both the current-detection technique and the gate-driving circuit allow to boost the power efficiency by 200% when compared to previous implementations of high-frequency neural stimulators [1], [2]. The circuit is implemented in a 0.18 μm HV CMOS process, and the total chip area is 3.65 mm 2 . Does the Evoked Cortical Response to Robotic Wrist Manipulations Differ Between Recoverers and Non-Recoverers after Stroke? Joost van Kordelaar, Mark van de Ruit, Teodoro Solis-Escalante, Leo Aerden, Erwin van Wegen, Alfred Schouten, Gert Kwakkel, Frans van der Helm Abstract: How much a stroke survivor will recover is still difficult to predict with current prediction models. Information about proprioceptive function may improve these models as it is assumed that proprioception is a prerequisite for regaining motor function. However, objective clinical tests to measure proprioception are currently lacking. Recent studies suggest that the signal-to-noise ratio (SNR) of evoked cortical responses to robotic wrist joint manipulations, as measured with electroencephalography (EEG), may objectively reflect the intactness of proprioceptive pathways to the cortex. The aim of this study was to assess whether this EEG-based metric differs between patients with high (‘recoverers’) and low (‘non-recoverers’) motor recovery. Data from 32 patients was used for the present study. Clinical examinations and EEG measurements took place within the first 3 weeks after stroke and at 5, 12 and 26 weeks. Patients were labelled as ‘recoverer’ (or ‘non-recoverer’) as their recovery on the upper limb section of the Fugl-Meyer Motor Assessment was larger (or smaller) than 50% of the maximal possible improvement on the assessment. EEG recordings were obtained in a customized measurement van with a 64 electrode cap while the wrist angle was manipulated with a haptic robot. Patients were instructed not to exert any force onto the robot (i.e. ‘relax task’). The cortical evoked response at electrode level was quantified by the SNR, which reflects the magnitude of the response relative to the magnitude of background EEG activity. Twenty-two patients were labelled as recoverers and ten as non-recoverers. In the recoverers, the SNR was -16.4 dB in the first and -16.3 dB in the last measurement. In the non-recoverers SNR slightly increased from -18.3 dB to -17.1 dB between the first and last measurements. None of the found changes within and between groups were significant (p > 0.05). The lack of significant differences between groups suggests that the SNR at electrode level does not distinguish between patients with high and low motor recovery after stroke. To improve our methodology we propose to adopt an EEG source localization method in order to identify only the sources that respond to the robotic manipulations and improve SNR specificity. Photoacoustic and High Frequency Ultrasound Imaging of Systemic Sclerosis Patients Khalid Daoudi, Brigit E. Kersten, Madelon C, Vonk, Chris L. de Korte Abstract: Systemic Sclerosis (SSc) is an autoimmune disease characterized by a triad of inflammation, vasculopathy, and fibrosis of the skin and internal organs such as gastrointestinal tract, heart, lungs, renal and kidneys. SSc can lead to premature death especially when there is pulmonary involvement. At early stages, alterations of blood vessel networks in the fingertip may occur in patients with SSc, leading to hypoxia as well as subtle skin thickening. Imaging these parameters could lead to early diagnosis of SSc patients and monitoring of early therapy responses. Currently, there is a need for a real-time and non-invasive imaging modality that combines morphological and functional imaging with the possibility to monitor the oxygen status of the blood vessel network. In this study, we investigated the feasibility to detect and diagnose systemic sclerosis by imaging microcirculation and oxygen saturation of blood vessels in nail-bed using photoacoustic imaging (PA) and estimating skin thickening using high-frequency ultrasound (HFUS). Thirty two subjects (both adult man and women) participated in this study: 12 patients with advanced systemic sclerosis, 5 patients with early systemic sclerosis, 5 primary Raynaud’s, and 10 healthy subjects as a control group. Nail bed oxygen saturation of patient with SSc whether it is in advanced or early stage showed a significant lower oxygenation (78.09 ±10.68 vs. 95.3 ±2.7, p < 0.0001) and a skin thickness difference (0.51 ±0.17 mm vs. 0.33 ±0.04 mm, p<0.005) compared to primary Raynaud’s and healthy volunteers. The PA and HFUS data was supported by conventional capillaroscopy imaging performed on all participants. This pilot study demonstrates the possibility to use photoacoustics and high-frequency ultrasound as a diagnostic tool for early detection of systemic sclerosis by imaging parameters not accessible with other imaging tool and might be used to non-invasively monitor early therapy response. The Design and Development of an Ambient Sensor System to Detect Incontinence Hannelore Strauven, Hans Hallez, Vero Vanden Abeele, Bart Vanrumste Abstract: Introduction. Over 50% of nursing home residents suffer from incontinence. Effort has been made to improve incontinence care management in nursing homes by measuring incontinence episodes and saturated incontinence material automatically. Although, most research uses sensors that are integrated in the incontinence material or attached to or implanted in the patient. Here, an unobtrusive alternative is presented. Materials and method. To detect urinary incontinence episodes in the air, off the shelf ammonia (NH3) gas sensors are implemented in the sensor system design. Sensors to detect ambient parameters such as temperature and humidity are integrated to improve the reliability of the sensor system. Through an ammonia dilution process, from 375 ppm to 12 ppm (i.e. in line with the typical NH3 concentration of urine), the possibility to detect urinary incontinence episodes in the air is validated. Results and discussion. The output signal of the gas sensors is proportional to the presented NH3 concentration in the dilution process. Furthermore, semiconductor gas sensors show a higher sensitivity than electrochemical cells for this application. The ambient parameter temperature strongly influences the gas sensors’ output signal. Conclusion. An ambient sensor system is presented that is able to detect NH3 up to concentrations comparable with the NH3 concentration in urine. Since the design is unobtrusive, it has little direct impact on the resident and fits in with other electronics present in the nursing home. Also, the system measures continuous and does not require extra actions from care personnel. Further research should include urine detection, sample preparation in incontinence material and measurements in the field. Machine Learning for Classification of Uterine Activity Outside Pregnancy Tom Bakkes, Federica Sammali, Nienke Kuijsters, Chiara Rabotti, Dick Schoot, Massimo Mischi Abstract: The increasing trend towards postponing childbirth has led to an increase in infertility risk [1]. Currently, around 1 in 5 couples have problems conceiving or maintaining an established pregnancy [2]. The hope of these couples often relies on assisted reproductive technologies, such as in-vitro fertilization (IVF). However, IVF failure rate is still above 70%. Although the causes for IVF failure are still not well understood, uterine receptivity seems a determinant factor for successful conception. Understanding the role of contractions outside pregnancy and possibly quantifying them may help understanding the link between normal uterine contractility and infertility, especially in relation to IVF failure. Recently, electrohysterography (EHG) and ultrasound (US) speckle tracking have shown to be promising for non-invasive, quantitative assessment of uterine activity. In this study, we investigated the use of machine learning for discriminating the uterine activity during the four phases of the menstrual cycle (menses, late follicle, early luteal, and late luteal phase). EHG and US acquisitions were performed on 6 healthy women in 4 phases of the menstrual cycle (total of 24 observations). A set of amplitude- and frequency-features were extracted from US and EHG data [3]. Four different types of classifiers were tested: support vector machine (SVM), K-nearest neighbours, Gaussian mixture model, and naïve Bayes. A full search was used to find the optimal combination of features and classifier parameters. Validation was performed by the leave-one-out method. According to the obtained results, the SVM classifier showed the best performance, resulting in an accuracy, sensitivity and specificity of 90%, 79% and 93%, respectively, by combining 2 US and 1 EHG feature. [1] A. A. de Graaff et al., Fertility and sterility( 95), 2011. [2] G. F. Whitman-Elia et al., J Am Board Fam Pract(14), 2001. [3] F. Sammali et al., Reprod Sci, 2018. [4] F. Sammali et al., IEEE Trans Ultrason Ferroelectr Freq Control, 2018 Neuro-Mechanical Control of Lower Limb Exoskeletons Guillaume Durandau, Herman van der Kooij, Massimo Sartori Abstract: One of the main difficulties in using exoskeletons for the rehabilitation of paretic patients is that we do not understand the neuro-mechanical interplay between wearable robots connected in parallel with the human body. This results in suboptimal rehabilitation solutions for the patient. To better understand this interplay, we first created a real-time electromyography (EMG)-driven model framework. This allows the computation of the different muscles states as well as joints moment. We showed the possibility of the EMG-driven model to work in real-time and to extrapolate outside of its calibration data. Following this encouraging result, we integrated our framework within a lower limb (knee and ankle) exoskeleton and used the computed biological joint moments to drive it. This class of controllers allowed the exoskeleton to be driven by healthy as well as paretic patients in sited experiments. Results showed that this approach can be used to alter EMG patterns consistently as a function of the level of assistance and reduce the normalized variability of the muscles EMG. Subsequently, we investigated the possibility to drive this model using a wearable inertial measurement unit (IMU) suit and compared it with conventional marker-based motion capture systems. This will enable us to conduct human-exoskeleton experiments outside of the laboratory. Our current work shows how by combining this IMU suit with a bilateral ankle exoskeleton, we can investigate the effect of the robot on the neuromusculoskeletal system during functional tasks like walking, stair and ramp. At the same time, we can use this knowledge to drive the exoskeleton to assist the user during these tasks. These different results demonstrated the possibility of these new developed tools to understand the neuro-mechanics of man-machine systems. This can be a game changer in bringing more efficient neurorehabilitation tool exploiting neuroplasticity. Human Behaviour Monitoring Through Smartphones Kostas Konsolakis, Hermie Hermens, Oresti Banos Abstract: The recent technological advances have enabled new potentials for monitoring people’s health. Human behaviour analysis through smartphones has been an active field for more than a decade, playing a major role in promoting active and healthy ageing. Smartphone sensors, such as accelerometer and gyroscope, but also GPS, Wi-Fi, and Bluetooth, can detect users’ activity and movement (for example sitting, walking, running) [1]. Data related to screen touch events and users’ response time can be used to track cognitive states, such as attention and alertness [2]. Phone calls and text messages have been used to monitor users’ social life [3], while audio and microphone signals, combined with the users’ physical activity, can be used to detect mood (boredom, happiness, anxiety, etc.) [4]. Even though human behaviour can be detected through physical, cognitive, emotional and social sensing, there is not a clear investigation for detecting human behaviour in a holistic approach, using smartphone data. The present study aims to investigate human behaviour through smartphone data, by answering the following research question “Could a smartphone monitor human behaviour in a broad sense?”. For the data acquisition phase, we will conduct an experiment in both a controlled and uncontrolled environment. Subjects will be asked to participate voluntarily and perform a series of consecutive tasks. Four main scenarios will be investigated by examining different classification approaches, related to users’ physical activities, social interaction, cognitive abilities and emotional states. Based on the collected data from the controlled experiment the model will be trained to predict momentary behaviours in a short-term period, while data from the uncontrolled environment will be used to detect behaviour routines and changes over a longer period of time. Blood Flow Patterns in the Left Ventricle with Low-Rate 2D and 3D Dynamic Contrast-Enhanced Ultrasound Peiran Chen, Ruud van Sloun, Simona Turco, Hessel Wijkstra, Patrick Houthuizen, Massimo Mischi Abstract: Heart-failure is one of the leading causes of death world-wide. In heart-failure, the main pumping chambers of the heart may become stiff, resulting in inefficient blood pumping. A clinically-proven treatment option for heart failure is cardiac resynchronization therapy (CRT), which aims at improving pumping efficiency. Functional impairment of the left ventricle (LV) commonly occurs already at the early stage in the development of heart disease. Since blood flow patterns are strongly associated with the ventricular function, cardiac diagnostics may benefit from flow-pattern analysis. Several blood flow-pattern visualization tools have been proposed that require ultrafast ultrasound acquisitions, while techniques that can deal with low frame rates are still lacking, especially for the emerging 3D ultrasound, for which the volume rate is intrinsically low. To overcome this limitation, we propose a novel technique for the estimation of LV blood velocity and relative-pressure fields from 2D dynamic contrast-enhanced ultrasound (DCE-US) at low frame rates. Different from other techniques, our method is based on the time-delays between time-intensity curves measured from a set of neighboring pixels in the DCE-US video. Using Navier-Stokes equation, we regularize the obtained velocity fields and derive relative-pressure estimates. Blood flow patterns are characterized in terms of their vorticity, changes in relative pressure (dp/dt) in the LV outflow tract, and viscous energy loss, as these reflect the ejection efficiency. Since the goal of CRT is improving pumping efficiency, we evaluated the proposed method on 18 patients (9 responders and 9 non-responders) who underwent CRT. After CRT, the responder group evidenced a significant (p<0.05) increase in vorticity and peak dp/dt, and a non-significant decrease in viscous energy loss. No significant difference was found in the non-responder group. Relative feature variation before and after CRT evidenced a significant difference (p<0.01) between responders and non-responders for all the features. In the future, the feasibility of the method for 3D DCE-US will be investigated. E-Supporter: Personalized Technology Supported Coaching of Patients with Chronic Diseases Wendy Oude Nijeweme – d’Hollosy, Laura Schrijver, Miriam Vollenbroek - Hutten Abstract: A serious challenge in current Western societies is the increasing number of patients with chronic diseases due to unhealthy lifestyle and the growing number of older people. Currently, more than 52% of the Dutch people have a chronic disease of which at least 33% have multiple diseases. Unfortunately, the number of people with a chronic disease is expected just to increase leading to a higher pressure on the quality of healthcare and rising healthcare costs. Personalised coaching to support patients’ self-efficacy on self-management could be a key solution to address these problems. In the last decade, innovative technologies have been developed that focus on self-management of patients. However, this support cannot be fully tailored to the patient as these apps usually do not take into account patients’ behaviour. Therefore, the development of e-supporter has been started in September 2018. E-supporter is an element of the Centre for Care Technology Research (CCTR) e-manager project that aims to improve the quality of life and perceived quality of care in patients with chronic conditions. The concept of e-supporter has been based on literature research, expert meetings and pilot studies. E-supporter will integrate various e-coaching apps that help patients with COPD, asthma and diabetes mellitus type 2 (DM2) to self-manage their disease. These apps have been developed by Maastricht University Medical Centre (MUMC+), TNO, University of Twente and commercial parties. E-supporter will promote pursuing a personalised treatment plan in daily life by using an intelligent interface that uses information from the ABCC tool, the various integrated e-coaching apps and daily health behaviour, thoughts and moods of the patient. The ABCC tool assesses the disease state, disease burden, and personalized goals that the patient has set together with the healthcare professional during consultation. Tailored coaching content will be developed based on the I-Change model and prediction models that inform the patient of expected consequences and benefits of their behaviour. In addition, e-supporter also helps healthcare professionals to be better informed about patients to enhance personalised care. This all will lead to healthcare that is more efficient, cost-effective, sustainable and of high quality. Robot Based Hand Model Validation Jinne Geelen Abstract: The decoding abilities for brain-computer interface (BCI) systems should be improved to allow for daily use by patients in a home environment. One area of interest for decoding is the sensorimotor cortex. Studying the representation of hand and finger movement on the cortex provides insight into the relationship between biomechanics and neurophysiology. An analysis with the recently developed musculoskeletal model of the hand can support this research. The anatomical parameters of this model are based on the data from one cadaver study and it has not yet been validated for general use. The main research question is: Can the hand model describe muscle activity during individual finger movements for a general population? The intrinsic and the extrinsic hand muscle activity will be measured with surface electromyography (sEMG). Two distinct tasks will be performed both with the help of the finger perturbation robot (FPR). For the first tasks, the participant holds the finger handles of the FPR with the thumb and one of the other fingers in a pinching posture. A trajectory on the screen has to be followed by moving the finger handles. The depicted path consists of an increasing force level (ramp) followed by a constant force level. The second tasks consist of force and position perturbation trials again performed with the FPR. The participants will be asked to keep their position or force constant while a perturbation is applied with the FPR. Predictions of muscle contractions, based on the model, are compared with electromyography (EMG) measurements in the forearm and hand. Our results provide an indication of the required optimization steps to create a valid general hand model. The first improvement that we want to implement is the scaling of the model. The scaling will enhance the prediction of the muscle activity during individual finger movements in a general population. Independent Component Analysis with Input From High Density Electrocorticography Grids Meron Vermaas, Jinne Geelen Abstract: Our understanding of the origin of measured brain signals should be extended to improve signal decoding used in brain-computer interfaces (BCI). Finger movements are a promising candidate to control electrocorticography (ECoG) based BCI. A recent functional MRI study demonstrated the somatotopic organization of finger representation on the sensorimotor cortex. Activity in the sensorimotor cortex detected with fMRI relates to activity in high-density ECoG grids. Thus, the activity patterns underlying ECoG potentials are expected to be related to individual finger movements. The goal of this study is to estimate the independent source activity related to individual finger movements somatotopically mapped on the sensorimotor cortex. The ECoG data of one human participant is obtained with a high-density grid (32 electrodes) above the sensorimotor cortex during a finger flexion task. Finger flexion is measured during the task with a data glove. An independent component analysis (ICA) is performed and the optimal number of components is investigated. The components retrieved from the ICA are correlated to the finger flexion data. Probability maps of the components are computed to check the stability of the ICA outcome. The spatial spread of the components over the HD grid is examined and compared to the expected activity pattern based on a volume conduction model. The successful application of ICA on HD ECoG grids in this exploratory study could be an indication for the feasibility of inverse modelling and source reconstruction. High-Volume-Rate 4D Echo-PIV in a Dynamic Left Ventricular Phantom Jason Voorneveld, Hicham Saaid, Christiaan Schinkel, Antonius van der Steen, Frank Gijsen, Nico de Jong, Tom Claessens, Sasa Kenjeres, Johan Bosch Abstract: Left ventricular (LV) blood flow is a promising early stage biomarker of ventricular dysfunction. However, blood flow in the LV is a 3D phenomenon and current clinical measurement techniques are limited to 1D /2D. Echo-particle image velocimetry (echo-PIV) can be used to measure 3D blood flow patterns but requires high frame rates to effectively track the fast flows expected in the LV. We use a prototype transesophageal (TEE) matrix probe, capable of 3D high volume rate (4 kHz) imaging and test the capabilities of 4D echo-PIV in a realistic and dynamic LV phantom. A compliant, optically and acoustically transparent silicone LV chamber, encased in an acrylic box, was fitted with bio-prosthetic mitral and aortic valves (Edwards), connecting to atrial and compliance chambers. The system uses a piston pump (ViVitro) to impose pressure and volume changes in the acrylic box, causing realistic flow patterns to circulate inside the LV. The ‘ground truth’ flow patterns in the LV are captured using tomographic PIV, a 3D optical velocimetry technique with very high spatial (12.4 pixel/mm) and temporal (2000 fps) resolution. Using the TEE matrix probe (Oldelft, 5 MHz), connected to a Verasonics Vantage 256, single diverging waves were transmitted to insonify a field of view of 24 x 24° up to a depth of 10 cm. Nine overlapping beams were acquired in a gated sequence with the pump cycle to obtain the full field of view of the LV without reducing frame rate. Ultrasound contrast agent (SonoVue, 20µl/l) was added for visualizing LV flow. Echo-PIV was then performed in the spherical domain using normalized cross correlation with a kernel size of 5.7mm x 12° x 12° and an overlap of 50% x 75% x 75%, resulting in a vector map resolution of 2.5mm x 3° x 3°. The vector maps obtained with echo-PIV were then qualitatively compared to the tomographic PIV results. The high velocity trans-mitral jet velocities of ~1 m/s could be captured by 4D-echo-PIV although there was still some underestimation present, especially in the lateral directions. This phantom is a useful tool for further optimization of echo-PIV. An MRI-Based Pipeline to Register Patient-Specific Wall Shear Stress Data to Histology Astrid Moerman, Suze-Anne Korteland, Stefan Klein, Kim van Gaalen, Kristina Dilba, Dirk Poot, Aad van de Lugt, Ellen Rouwet, Jolanda Wentzel, Anton van der Steen, Frank Gijsen, Kim van der Heiden Abstract: Wall shear stress (WSS), the frictional force exerted on endothelial cells by blood flow, is hypothesized to influence atherosclerotic plaque growth and composition. We developed a pipeline for image registration of MR and histology images of advanced human carotid plaques and corresponding WSS data. The pipeline required four types of input images, in vivo MRI, ex vivo MRI, photographs of transversal sectioned plaque tissue (en face images) and histology images. These images are transformed to a shared 3D space by applying a combination of rigid and non-rigid registration algorithms. Transformation matrices obtained from registration of these images were used to transform subject-specific WSS data to the shared 3D space as well (see figure). WSS values originating from the 3D WSS map were projected in 2D on the corresponding lumen locations in the histological sections and divided into eight radial segments. In each radial segment, the correlation between WSS values and histological parameter could be assessed. The pipeline was successfully applied to two carotid endarterectomy specimen. The resulting matched contours from the histology and WSS data were well aligned with a Hounsfield distance between 0.55 and 0.75 mm, which is below the in vivo MRI resolution. We investigated the robustness of the image registration pipeline by relocating the WSS data with respect to the stack of histology images, simulating the effect of a mismatch in the rigid registration of imaging modalities on WSS results for 184 segments. This relocation altered the mean WSS values projected on radial bins by 0.2 Pa, compared to the output of optimal registration. Moreover, a mismatch of one image slices changed correlation coefficient between WSS and plaque thickness by l0%, and never lead to a change in sign of the correlation. We conclude that the created pipeline offers the unique opportunity to robustly and quantitatively determine whether correlations between WSS and plaque composition exist. “Sprint Splint” – a 3D Printed, Rapidly Customized, Patient Specific, Wrist Splint Marit HN van Velzen, Merel W Eggenkamp, Samantha de Graaf, Quinten J Mank, Twan Simons, Arjo J Loeve, Alina G van der Giessen Abstract: INTRODUCTION Carpal tunnel syndrome (CTS) is amongst the most common upper extremity pathologies. Conventional therapy consists of fitting a wrist splint that prevents wrist flexion-extension. Unfortunately, the most-used splint type quickly gets dirty, while it cannot (easily) be cleaned. Furthermore, these splints completely cover the wrist and large parts of the hand and forearm, causing sweating, itching and skin irritation. Consequently, there is low therapy compliance in patients needing the splint, despite the fact that bracing is such a simple way to effectively and non-invasively heal CTS patients. The aim of the current work was to develop the “Sprint Splint”: a system to easily and rapidly make patient specific wrist splints through 3D printing without the need for time-consuming scans and adjustments, for better therapy compliance. METHODS Splint concepts were made based on functional, design and comfort requirements. The main functional requirement was that backlash of the wrist and finger-joints should not exceed those in conventional splints. Regarding comfort: no pressure points, no skin damage and painless splint donning and doffing. Additionally, the skin had to stay uncovered as much as possible without losing any functionality. Through Harris Profiles and interviews with expert users a final design was selected and tested. RESULTS A flat-printed, base-model splint was set up in FreeCAD-software. The base-model splint consisted of a flat-printable design of the splint, including an adjustable thumb opening and hexagon plane filling structure. To personalize the base-model, a total of 10 easy-to-measure dimensions of the patient’s hand, wrist and forearm are entered in a table linked to the base-model, directly returning a patient specific splint. After 3D printing the splint, it is finalized by molding it around the wrist of the patient. Six volunteers received well-fitting splints based on the workflow described above. CONCLUSION A method comprising a parametrized base-model splint, a dimensions input table and simple 3D printing and molding steps was developed to deliver patient specific 3D printable wrist splints. The method uses 10 easily measurable dimensions for customization. After further improvements, the Sprint Splint may offer a rapidly customizable, comfortable, cleanable and cheap alternative to current CTS splints. Screening of Foot-Related Musculoskeletal Problems in Non-Sedentary Employees. How can We Improve the Quantification? Kris Cuppens, Luiza Muraru, Tom Saey, Eveline De Raeve, Mario Broeckx, Veerle Creylman Abstract: Due to prolonged walking and standing, employees who have a non-sedentary job have an increased risk of developing musculoskeletal complaints related to foot and ankle, knee, hip and (lower) back. In the prevention of health risks in these employees, most attention is given to maintaining a correct posture and using the correct lifting techniques. Less attention is given to identify foot related problems in an early stage. To help companies that provide prevention services in screening employees on foot related problems, a screening tool was developed. In this tool, the employee fills in questions related to his personal health and the working environment. In a second stage, the occupational physician uses the tool to examine the employee on e.g. foot deformities, pain, callus formation and gait pattern. Based on the examination, the tool generates advice on e.g. footwear, performing stretching/strengthening exercises or the employee is referred to a foot specialist. As an occupational physician is not always specialized in examining the foot, we investigated the use of sensors to assist him/her in doing the examination. In this way the measurements are quantified, which helps to objectively follow up the employees on the long term, and keep track of their evolution. In our setup, specific examinations of the screening tool are linked to a specific measurement with a dedicated set of sensors. The possibility of using specific sets of sensors is evaluated in a feasibility study. More specifically, in-shoe pressure sensors are used to quantify the work environment (identifying prolonged periods of walking and standing, lifting loads, climbing stairs), accelerometers attached to the foot are used to measure the flexibility of the foot joints, and low-cost 3D scanners are used to identify deformations of the foot. In the feasibility study, 8 subjects are included, of which 6 wear orthopaedic insoles. The tests demonstrate the feasibility of measuring the subjects with these non-intrusive sensors and the added value for the examination of the subject. However, a more extensive validation is needed to determine the accuracy of the proposed setup. Age and Gender Effects on Resting State Networks after Mild Traumatic Brain Injury Mayra Bittencourt-Villalpando, Harm J. van der Horn, Edith Liemburg, Natasha M. Maurits, Joukje van der Naalt Abstract: Introduction: Cognitive complaints are common within the first weeks after mild traumatic brain injury (mTBI) and may persist for months to years in a subgroup (≈20%). Age-related cognitive decline can worsen these symptoms. However, effects of age on mTBI sequela have scarcely been investigated. Methods: Fifty-four patients (mean age: 37 years, range 19-64 years, 67% male) and twenty healthy controls (HCs) (mean age: 36 years, range: 18-61 years, 70% male), group-matched for age and gender were included. Functional magnetic resonance imaging (fMRI) was performed during 10 minutes rest. Independent component analysis (ICA) was used to identify resting state networks (RSNs). A multivariate approach was executed to evaluate the effects of age and gender on the RSNs in terms of the three main outcome measures: a) spatial map intensity, b) time-course spectral power, and c) functional network connectivity. Results: Age-related effects were identified for all three measures, indicating a) significant intensity decreases for increasing age in the default mode network (DMN), b) significant power spectral decrease within the 0-0.15 Hz range for increasing age in the DMN, frontoparietal network (FPN) and dorsal attention network (DAN), which is consistent with previous studies on healthy participants, and c) significant power spectral increase within the 0.15-0.25 Hz range for increasing age in the DMN, FPN and DAN. Additionally, a significant connectivity increase between the orbitofrontal and language networks was found for increasing age. Gender effects of smaller magnitude were found on spatial maps, with greater intensity found for males in the posterior part of the DMN, specifically within the cuneus. The significance threshold was set at p<0.05, with correction for multiple comparisons. Conclusions: This study on mTBI patients and healthy controls showed age- and gender-related effects on brain networks that have previously been suggested to be involved in cognitive functioning. How these effects are related to patients’ cognitive functioning in comparison to the healthy control group is still not clear and requires further investigation. Microelectrode Array (MEA) Measurements From Human Induced Pluripontent Stem Cell-Derived Neural Cultures for Psychiatric Disorders Areti Sfakianou, Femke de Vrij, Steven Kushner, Virgilio Valente Abstract: The pathophysiology of many neurological and psychiatric diseases remains undiscovered, due to our limited understanding of the biological mechanisms underlying these disorders. Induced pluripotent stem cell (iPSC) technology provides the unique opportunity to study neural cell cultures of individual patients in-vitro [1], thus enabling to assess the physiology of psychiatric disorders. Several protocols used to obtain iPSCs-derived neuron subtypes, networks or whole brain organoids, expose low efficiency, often resulting into immature neuronal cultures. In addition, the conventional adoption of co-cultures, especially with exogenous mouse astrocytes, increases variability. Growing human iPSCs-derived neural cultures from a common neural progenitor cell (NPC), has shown to improve the differentiation efficiency [1]. NPCs are differentiated into functional neural network cultures of neurons and astrocytes in a controlled ratio, without the need of exogenous astrocyte co-culture [2]. Monitoring of the differentiation process is conventionally performed by patch clamp recordings. Single-cell patch clamp measurements, however, do not reveal network behaviour during the differentiation process, which is a critical aspect in potentially assessing the biological mechanisms of psychiatric disorders [1]. In this work, we combine the simplified neural differentiation protocol with the use of microelectrode arrays to record and stimulate neural activity at the network level during differentiation period using a MEA system (Multi Channel Systems) . Measured cell culture data is analysed using a commercial analysis software (Multiwell-Analyzer), to assess spontaneous and stimulated network activity, synchronicity and bursting activity during a developmental phase. Long-term characterisation of human iPSC-derived neural network cultures using MEA recordings helps us gain more detailed knowledge for the biological mechanisms that underlie neuropsychiatric disorders, for phenotype screening and for the development of personalized treatment and drugs with medium-throughput capacity. References [1] N Gunhanlar, G Shpak, M van der Kroeg, LA Gouty-Colomer, ST Munshi, B Lendemeijer, M Ghazvini, C Dupont, WJG Hoogendijk, J Gribnau, FMS de Vrij and SA Kushner. A simplified protocol for differentiation of electrophysiologically mature neuronal networks from human induced pluripotent stem cells. Molecular Psychiatry, 23, 1336-1344. doi:10.1038/mp.2017.56 (2018). [2] Frega, M., van Gestel, S. H., Linda, K., van der Raadt, J., Keller, J., Van Rhijn, J. R., Schubert, D., Albers, C. A., Nadif Kasri, N. Rapid Neuronal Differentiation of Induced Pluripotent Stem Cells for Measuring Network Activity on Micro-electrode Arrays. J. Vis. Exp. (119), e54900, doi:10.3791/54900 (2017). Co-Integration of Flip-Tip Patch Clamp and Microelectrode Arrays for In-Vitro Recording of Electrical Activity of Heart Cells Asli Yelkenci, Ronaldo Martins Da Ponte, Virgilio Valente Abstract: The patch clamp has been widely considered the gold standard to measure intracellular ionic activity of single cells [1]. However, patch clamping is a laborious method and suffers from low throughput. To mitigate the disadvantages of patch clamping, planar patch clamp (PPC) chips with higher throughput have been recently introduced [2-3]. Yet those microfluidic chips do not allow to concurrently monitor the extracellular and the intracellular activity of the cells. Understanding of the complex cellular network activity and electrochemical processes, requires correlation between local field potentials (LFPs) of a population of cells and action potentials (APs) of single cells. This abstract presents a novel CMOS compatible microfluidic system that integrates flip-tip planar patch clamps (FTPPCs) and microelectrode arrays (MEAs) on the same wafer, for invitro extra- and intra-cellular recordings of electrical activity of cardiac cells. The device is fabricated using conventional wafer front- and back-side photolithography. The fabrication process leverages anisotropic wet etching selectivity of potassium hydroxide (KOH) and deep reactive ion etching (DRIE) to pattern FTPPCs. Before DRIE process, plasma-enhanced chemical vapor deposition (PECVD) of silicon dioxide (SiO2) is applied as passivation layer. After DRIE process, a metallization step is performed by sputtering titanium nitride (TiN) on patterned structures. As the final step, SiO2 is removed and backside DRIE is used to open apertures approximately with 2 µm diameter. The FTPPCs are intended to have a tip in 20 µm depth after KOH etching, and a spacing of 200 µm to ensure that mechanical stability of the device after DRIE. The planar MEAs are then patterned on the front side with 50 µm diameter and a pitch of 200 µm. A PDMS culture chamber is attached the front-side of the wafer, while a PDMS microfluidic channel is constructed on the back-side. By applying suction through the microfluidic channels, the cells are trapped in the FTPPC apertures. Potentiostatic measurements are used to record the ionic activity of the cells intracellularly, while low-noise instrumentation amplifiers are used in combination with the MEAs, to concurrently measure LPFs. Co-integration of PPC and MEAs on the same wafer can provide valuable insight in the correlation between single-cell activity and cellular network dynamics of heart cells in healthy and pathological states. Ultrasound Image Analysis to Quantify Needle Visibility and Echogenicity Nick van de Berg, John van den Dobbelsteen Abstract: Approximately 70% of interventional radiologists experience difficulties in percutaneous lesion targeting as a result of poor needle visibility in ultrasound images [1]. Visibility may be limited by acute needle insertion angles, inadequate image plane alignments, echogenic surrounding tissues, and imaging artefacts, such as reverberations, comet tails, or shadowing effects. Numerous techniques have been proposed to enhance needle visibility by either changing the imaging method or the needle itself, e.g. beam steering, spatial compound imaging, echogenic coatings, and surface scoring or denting. Needle visibility is typically addressed by observations and subjective ratings. In contrast, this work proposes an objective and methodical quantification method to determine and compare the efficacy of needle visibility enhancement techniques. An automated image acquisition and analysis approach is proposed to efficiently collect and summarize large amounts of needle visibility data. Image processing included filtering, line fitting and image intensity analyses. A contrast-to-noise ratio and signal ratio were used to measure needle visibility and echogenicity, respectively. The approach was evaluated in PVA, in a comparative study of commercially available needles, as well as custom-made needles with grooves that resemble compliant joints of steerable needles. Distinct visibility peaks were seen at the 90° needle insertion angle, when specular reflections returned to the US transducer. Bevelled (chiba) and diamond (trocar) needle tips had a high visibility for a large angular range, compared to conical tips. The highest needle shaft visibility resulted from grooved needles, followed by trocar and shiba needles, respectively. Tip and shaft visibility in ultrasound images were compared for a near-full range (25-335°) of possible needle insertion angles. The shafts with grooves and the tips of trocar and chiba needles had the best performance. Overall, the presented approach provided an efficient and objective method to evaluate and compare needle visibility enhancement techniques. 1. de Jong, T.L., et al., Needle placement errors: do we need steerable needles in interventional radiology? Medical Devices-Evidence and Research, 2018. 11: p. 259-265. Design, Development and Application of a Miniature Implantable Electronic Stimulator Ernest Boskovic, Marcel G.J. Nederhoff, Dyan Ramekers, Leonard J. van Schelven Abstract: DESIGN, DEVELOPMENT AND APPLICATION OF A MINIATURE IMPLANTABLE ELECTRONIC STIMULATOR Ernest Boskovic*, Marcel G.J. Nederhoff, Dyan Ramekers and Leonard J. van Schelven Department of Medical Technology & Clinical Physics, University Medical Center Utrecht, Heidelberglaan 100, 3508 GA Utrecht, The Netherlands ABSTRACT To conduct specific scientific research in small animals, including a chronic electrical stimulation, researchers need a sufficiently small implantable device that can apply these electric stimuli, and which uses a study dependent stimulation protocol. The total desired stimulation period can vary from several weeks to even a few months. We designed, developed and manufactured such a miniature low-power programmable stimulator and successfully applied it in two studies. We made two versions, a very small stimulator for mice and rats (25x14x5mm) and a slightly bigger stimulator for guinea pigs (32x23x6mm). The stimulator can be programmed to deliver a specific pattern of controlled electrical current pulses. The elliptically shaped enclosure consists of a waterproof resin cast with a thin silicone skin for biocompatibility. The stimulator has two output wires covered by biocompatible silicone tubes for electrical isolation. It is powered by a non-rechargeable battery. The current output has a compliance voltage of about 10 V and can deliver several mA of monophasic pulses, with a repolarization phase for discharge purposes. The maximal achievable duration is dependent on the stimulation protocol used. After implantation of the device, under the skin of the animal, communication is possible using an infrared remote control. This allows turning the device on or off, reading its serial number or adjusting stimulation protocol settings. Thanks to this novel device, our researchers are already able to conduct two studies, one into the effect of the vagus nerve stimulation on atherosclerosis in mice and rats, and a second study on electrical stimulation in combination with specific growth factors of mostly degenerated auditory nerve fibers in an attempt to functionally preserve those hearing cells in guinea pig. Number of ECG Replicates and QT Correction Formula Influences the Estimated Qt Prolonging Effect of Novel Drugs Hein van der Wall, Pim Gal, Koos Burggraaf, Michiel Kemme, Gerard van Westen Abstract: QT interval assessment is a pivotal step in the development of novel compounds, but the thorough QT study is under debate. The present analysis evaluated the effect of the number of short term ECG replicates extracted from a continuous ECG recording on the assessment of the QT interval and the effect of different QT correction formulas. Aims The aim of the present analysis was to demonstrate the feasibility of a novel approach in which several epochs extracted from a continuous ECG recording were used to assess the compound’s effect on the QT interval. The optimal number of ECG epochs (replicates) required to asses this effect were investigated. In addition, different QT correction formulas were compared. Methods For one hundred healthy volunteers, who received a compound prolonging the QT interval, 18 ECG replicates within a 3 minute window were extracted from 12-lead Holter ECGs. Ten QT correction formulas were deployed and the QTc interval was controlled for baseline and placebo and averaged per dose level. Results / Conclusions The mean prolongation difference was >4 ms for single and > 2 ms for triplicate ECG measurements compared to the 18 ECG replicate mean value. The difference was <0.5ms (10 % of the safety limit) after 14 replicates. In contrast, concentration-effect analysis was independent of both the replicate number and correction formula. Auxetic Meta-Biomaterials -A Mechanical and Morphological Assessment Helena Kolken, Karel Lietaert, Tom van der Sloten, Behdad Pouran, Amir Zadpoor Abstract: Total joint replacements are often referred to as one of the most successful surgical interventions, but in the light of current developments, young and active patients will most certainly outlive their implants. The mechanical failure of the bone-implant interface is one of the main reasons. With the introduction of a metallic implant, external loads are no longer carried by the bone alone. Instead, most of the load is carried by the implant, which will cause bone to adapt itself (Wolff’s Law) by reducing its volume in places it is no longer needed. Biomaterial optimization is therefore inevitable, when working towards the next generation of life-lasting implants. The emerging concept of metamaterials offers a promising route to the development of such implant biomaterials with unique combinations of mechanical (e.g. Negative Poisson’s Ratio), mass-transport (e.g. permeability) and biological properties (e.g. tissue regeneration performance). The topology of these so-called meta-biomaterials, may be rationally designed to exhibit unprecedented properties for tissue regeneration and sustained mechanical support.1 Unlike conventional meta-biomaterials, auxetic meta-biomaterials have a negative Poisson’s ratio and expand laterally in response to axial stretch.2 A recent study has proven their importance within the field of orthopedics, by improving implant-bone contact and potentially implant longevity in the hip stem.1 Laterally applying an auxetic meta-biomaterial resulted in compression on both of the implant’s contact lines with the surrounding bone, decreasing the chance of bone-implant interface failure. In this work, we characterize the mechanical properties of additively manufactured, Ti-6Al-4V auxetic lattices, based on the re-entrant hexagonal honeycomb. The mechanical properties of this unit cell can be tuned through slight alterations in its geometry, to obtain a variety of Poisson’s ratios. The limits of the Selective Laser Melting (SLM) process were explored to synthesize structures with optimal bone-ingrowth properties. Their architecture was evaluated using micro-CT, while its mechanical properties were assessed under compression with the help of Digital Image Correlation (DIC). With this comprehensive library of mechanical and morphological properties, we hope to contribute to the adoption of auxetic lattices as ideal substitutes for bone in life-lasting implants. An Angle-Independent Cross-Sectional Doppler Method for Flow Estimation in the Common Carotid Artery Luuk van Knippenberg, Ruud van Sloun, Arthur Bouwman, Massimo Mischi Abstract: Doppler ultrasound is an important technique for non-invasive quantification of blood flow, which is of major clinical importance in the assessment of cardiovascular condition. However, a major disadvantage of flow estimation using Doppler ultrasound is the operator-dependency in achieving a longitudinal image in which both the Doppler angle (beam-to-flow angle α) and vessel diameter (flow area) can be accurately estimated. Instead, keeping the probe correctly positioned in the short axis is much easier and has the advantage of capturing the whole flow profile of a vessel while measuring the vessel area simultaneously [1]. Therefore, cross-sectional imaging would be preferable to longitudinal imaging if α can be estimated. In this work, we propose a solution that is based on modelling of the common carotid artery as a cylinder, which intersects with the ultrasound plane, resulting in an ellipse on the B-mode image. The ellipse characteristics (semi-major axis a, semi-minor axis b and tilt β) are used to estimate α by solving a least-squares problem. To prove theoretical feasibility a geometric model was simulated in MATLAB, where both the cylinder (r=5 mm) and imaging plane can be arbitrarily rotated and α is estimated from the resulting intersection. In addition, the propagation of erroneous estimates of b (5% error, 0.25 mm) was evaluated. Finally, blood flow through a tilted and rotated vessel was simulated using Field II software [2][3]. In the geometrical model, α was estimated perfectly (zero error with respect to reference), although the sign of the Doppler angle could not be determined (i.e. 60° vs 120°). For a 5% error introduced in b, the error in estimating α increased as α approached 90°, from 1.6° at a Doppler angle of 30° up to 8.0° at 70°. In the Field II simulation (α=135° and r=5 mm), the estimates were αest=132.6° and rest=4.9mm. The proposed method to estimate the Doppler angle from cross-sectional ultrasound images shows promise and satisfactory results in these theoretical experiments. However, further research should be performed to evaluate how sensitive this method is to errors in the estimated ellipse parameters and how this method performs in vivo. References [1] R. Schorer, A. Badoual, B. Bastide, A. Vandebrouck, M. Licker, and D. Sage, “A feasability study of color flow doppler vectorization for automated blood flow monitoring,” J. Clin. Monit. Comput., vol. 31, no. 6, pp. 1167–1175, 2017. [2] J. A. Jensen, D.- Lyngby, P. Medical, B. Engineering, J. A. Jensen, and I. Technology, “Field: A Program for Simulating Ultrasound Systems,” Pap. Present. 10th Nord. Conf. Biomed. Imaging Publ. Med. Biol. Eng. Comput., vol. 34, pp. 351–353, 1996. [3] J. A. Jensen and N. B. Svendsen, “Calculation of pressure fields from arbitrarily shaped, apodized, and excited ultrasound transducers,” IEEE Trans. Ultrason. Ferroelec., Freq. Contr., vol. 39, no. 2, pp. 262–267, 1992. Towards a Semi-Flexible Parylene-Based Platform Technology for Active Implantable Medical Devices Nasim Bakhshaee Babaroud, Marta Kluba, Ronald Dekker, Wouter Serdijn, Vasiliki Giagka Abstract: Active implantable medical devices have been developed for diagnosis, monitoring and treatment of large variety of neural disorders. Since the mechanical properties of these devices need to be matched to the tissue, soft materials, such as polymers are often preferred as a substrate [1]. Parylene is a good candidate, as it is highly biocompatible and it can be deposited/etched using standard Integrated Circuit (IC) fabrication methods/processes. Further, the implantable devices should be smart, a goal that can be accomplished by including ICs. These ICs, often come in the form of additional pre-packaged components that are assembled on the implant in a heterogenous process. Such a hybrid integration, however, does not allow for size minimization, which is so critical in these applications, as otherwise the implants can cause severe damage to the tissue. On the other hand, it is essential that all components are properly packaged to prevent early failure due to moisture penetration [2]. In this work we use a previously developed semi-flexible platform technology based on a Parylene substrate and Pt metallization, which allows integration of electronic components with a flexible substrate in a monolithic process. We use an IC fabrication-based platform that allows for the fabrication of several rigid regions including Application-Specific Integrated Circuits (ASICs) and other components connected to each other by means of flexible interconnects. We aim to add more functionality to this technology and thereby extend it to a platform for a variety of medical applications. An example of such functionality is integrating Light Emitting Diodes (LEDs) for optogenetic stimulation or integrating Capacitive Micromachined Ultrasound Transducers (CMUTs) for ultrasound stimulation or ultrasound wireless power transfer. Since the long-term reliability is critical for implantable devices, we intend to reinforce our implant with an extra Polydimethylsiloxane (PDMS) encapsulation layer that relies on the low viscosity of the uncured rubber to flow in every detail of the surface to prevent void formation [3]. Therefore, this work also focuses on enhancing the adhesion of PDMS to Parylene, as it must remain strong for the required lifetime of the device. The Influence of Soft Encapsulation Materials on the Wireless Power Transfer Links Efficiency Anastasios Malissovas, Wouter A. Serdjin, Vasiliki Giagka Abstract: As the era of Bioelectronic medicines (BEms) evolves, new technological challenges are generated, including miniaturized devices that are encapsulated with flexible materials and energized by wireless power transmission (WPT) techniques. Among them, magnetostatic, also known as inductive, and ultrasound (US) are the most viable candidates for shallow and deep applications, respectively. However, the conductive nature of the human tissue with high relative permittivity increases the parasitic components of the printed spiral coils (PSCs), while the acoustic impedance mismatch between the tissue and ultrasound transducers leads to power losses in the WPT link. This study focuses on the influence of biocompatible, soft, polymeric materials, such as polydimethyloxane (PDMS) and Parylene-C, on the electrical behaviour of the aforementioned externally powered receivers. Unlike previous works, this investigation includes the high gas permeability property of polymers, predicting the electrical impact of moisture absorption. Analytical and simulation models are utilized to discriminate the effect of various packaging schemes and to relate their influence on the WPT link efficiency. Lastly, empirical measurements in air and saline aim to verify the proposed methods. Early modelling results demonstrate that when a PSC is encapsulated with 50 μm PDMS and submerged into saline, its resonance frequency and quality factor are decreased by 3.6% and 34.2%, respectively. That renders the maximum theoretical WPT link efficiency to be reduced by 9%, compared to free-space propagation in air. Interestingly, when the coating thickness increases to 500μm, the WPT link efficiency drops only by 2.4%. In the case of US, similar effects are predicted, yet the influence of the coating materials will be different. More specifically, their acoustic impedance decreases the US transducers’ natural frequency of vibration and mechanical quality factor, due to the effect of added mass. In addition, when the coating thickness increases towards the wavelength of the incident US wave inside the material, the aforementioned effects become more evident. The outcome of this study aims to address the contributing factors on the WPT link power losses from the electronics packaging perspective and to suggest on how the effect of the surrounding medium could be mitigated, improving the WPT link efficiency. Relevance Ranking of Multi-Parametric Radiomics in Low-Dose CT for Discrimination Between Emphysematous and Non-Emphysematous Lung Tissue. Yeshaswini Nagaraj, Jiapan Guo, Monique D. Dorrius, Mieneke Rook, Qiong Li, Rozemarijn Vliegenthart, Matthijs Oudkerk, Peter . M.A van Ooijen Abstract: Emphysema is part of the chronic obstructive pulmonary disease (COPD) spectrum and known for its high prevalence and mortality rate worldwide. The appearing lung tissue destruction can be detected on Low-Dose Computed Tomography (LDCT) as decreased lung density and architectural changes. However, the latter is difficult to quantify so far. In this study, generalised matrix learning vector quantisation (GMLVQ) classifier was employed to discriminate between emphysematous and non-emphysematous lung tissue in LDCT by using radiomics features. 64 LDCT scans were randomly selected from a large lung cancer screening trial. Independently, three radiologists selected a total of 419 free form Regions of Interest (ROIs) from these scans. Every ROI consisted of three consecutive non-overlapping 2D slices. 300 ROIs were classified as emphysematous regions by the radiologists. During the multi-randomised training, each time 100 emphysematous regions were compared to 100 non-emphysematous regions. ROIs were used to extract radiomics features, which were ranked based on relevance factors from GMLVQ. We used feature selection techniques to decrease redundancy between features before performing the classification. Cross-validation was done using leave-one out procedure for each training set and number of feature selected. We extracted 1008 radiomics features including Gray-Level Co-Occurrence Matrix (GLCM), Gray-Level Run-length Matrix (GLRL), first order histogram features, shape descriptors and image filters. The GMLVQ classifier with the relevance feature ranking, which can be related to pairwise correlations between features, resulted in a maximal area under the curve of 0.95 for discrimination between emphysematous and non-emphysematous lung tissue. GMLVQ using radiomic features can discriminate between lung tissue with and without emphysema. Potentially, it can be a valuable tool for characterization of emphysema. Effects of Material Models on Micro-Motion Finite Element Analyses of Tibial Components in Primary Total Knee Arthroplasty Mauricio Saldivar, Thomas Anijs, Dennis Janssen, Nico Verdonschot Abstract: In cementless total knee arthroplasty (TKA) high micro-motions at the implant-bone interface are found to prevent local bone ingrowth, which can result in early aseptic implant loosening. In this study, micro-motions underneath a primary tibial tray are computed using the finite element method (FEM). Representation of intra-surgical bone damage is one of the factors which should be taken into account to simulate realistic micro-motion conditions. METHODS: A tibial FEM model was used to compare influence of four material models; a linear elastic model, a softening Von Mises (VM) criterion model, and respectively an ideal and a hardening isotropic crushable foam plastic model (CFM). Geometry and material properties were based on a calibrated CT-scan; the resulting tibia was virtually resected and implanted with a trabecular metal tibial component. An implantation test was performed to predict intra-surgical damage. Loading conditions representing consecutive cycles of walking, sitting, stair climbing and cycling were applied to estimate micro-motions during daily living. A micro-motion threshold of 40 μm is used to determine the extent of local bone ingrowth. RESULTS SECTION: The softening VM criterion model was the plastic material model in which the highest volume of excessive deformation, was found, followed by the hardening and ideal isotropic CFM respectively. Ingrowth distributions underneath the baseplate of the LE model was in line with both CFM’s, while less extensive in the VM model. Distinct differences between the LE and plastic models were found around the pegs, since relative LE ingrowth was very low. DISCUSSION: Current findings stress the importance of using an adequate plastic material model and inclusion of the implantation process when considering initial implant stability in TKA. When interpreting the extent and distribution of ingrowth, it is important to keep in mind non-implant specific native knee loads and an assumed binary micro-motion threshold of 40 μm are used. Also other mechanical stimuli may be involved in primary stability of press-fit implants, as increased strains are also linked to ossification. SIGNIFICANCE: Realistic simulations of peri-prosthetic micro-motions and related bone ingrowth can prevent cases of implant loosening and improves analyses of subsequent long-term bone remodeling in cementless implants. Towards Deployable Meta-Implants Françoise Bobbert, Shahram Janbaz, Amir Zadpoor Abstract: Meta-biomaterials exhibit unprecedented or rare combinations of properties not usually found in nature. Such unconventional mechanical, mass transport, and biological properties could be used to develop novel categories of orthopedic implants fulfilling desired requirements, otherwise known as meta-implants. In this study, we used bi-stable elements working on the basis of snap-through instability to design deployable meta-implants. Deployable meta-implants are compact in their retracted state, which enables the possibility to bring them to the surgical site with minimum invasiveness. Once in place, they could be deployed into their full-size load-bearing shape. By using minimally invasive surgery, the recovery time of the patient and the risk of infection could be reduced. We used an Ultimaker 3D printer to manufacture biocompatible PLA bi-stable elements. Five types of meta-implants were created by arranging these bi-stable elements in such a way to obtain a radially deployable structure, three types of auxetic structures, and an axially deployable structure. The intermediate stable conditions (i.e. multi-stability features), deployment force, and stiffness of the meta-implants were found to be strongly dependent on the geometrical parameters of the bi-stable elements as well as on their arrangement. The high porosity of these deployable structures allows for improved bone ingrowth. Because multi-stable deployable impl¬¬ants have different stable equilibrium states, a major design challenge is to ensure that they provide enough mechanical support in their deployed configuration. Future research should therefore be focused on evaluation of the mechanical performance of meta-implants as well as on designing miniaturized versions that make them more suitable for the application as bone substitutes. Towards an Active Graphene-PDMS Implant Gandhika K. Wardhana, Wouter A. Serdijn, Sten Vollebregt, Vasiliki Giagka Abstract: Neural interface in the form of microelectrodes are used to monitor and treat spinal cord injury and other neurological disorders by the means of recording and stimulation. Despite of the apparent result of these electrical interventions, understanding of the mechanism behind neural stimulation is still inadequate. The use of optical monitoring during implantation is limited due to the use of opaque electrode partially blocking the implantation site. While the use of transparent conductor for electrode is not uncommon in general electronics where indium tin oxide (ITO) is widely used for displays, however ITO is not suitable for implantation due to its brittle nature[1]. An alternative material to fabricate transparent electrodes is graphene, a single layer of carbon atom forming sp2 hybridization. Its high charge mobility, flexibility, mechanical strength and optical transparency make it suitable for various flexible electronics applications including implantable microelectrode arrays. In biomedical fields, graphene has shown potential application as biosensor, stimulation and recording electrode[2]. Although fabrication of graphene microelectrodes has been previously shown[3], graphene had to be transferred manually for each individual implant. The high temperature needed during graphene deposition makes device fabrication directly on the flexible material impossible. Instead, the fabrication process relies on a transferring process of graphene layer from growing medium with high thermal budget to another desired substrate. Manual transfer process of graphene is a skill-dependant process with low scalability. In this work, a method of fabricating encapsulated graphene electrodes in polydimethylsiloxane (PDMS) with a controlled wafer-scale graphene transfer is proposed. Graphene transfer is done by wafer-assisted PDMS-PDMS bonding to minimalize operator dependency. The novel use of PDMS as encapsulation material for graphene electrode is due to its biocompatibility, flexibility and optical transmittance. Difference in material characteristics, such as the thermal expansion coefficient has become one of the challenges during fabrication process. Despite of these challenges, the prospect of transparent implant has been shown in preliminary testing on optical transmittance of graphene layer on PDMS with up to 77% transmittance in the visible light spectrum. While full characterization of the device is still in progress, further results will be reported during the conference. Signal Specific SAR ADC for Multi-Channel Atrial Electrogram Signal Acquisition Samprajani Rout, Samaneh Babayan Mashhadi, Wouter A. Serdijn Abstract: Atrial electrograms (AEGs) are the signals recorded on the surface of the heart and can be used to study the signal propagation in the atria with an aim to understand the development and progression of atrial fibrillation, which is a type of cardiac arrhythmia. AEGs are acquired using a high resolution electrode array which pose strict constraints on the power and area consumption of the data acquisition system. For a single channel, a dedicated analog-to-digital converter (ADC) is used. However, this approach becomes hardware intensive as the number of channels increase. Multi-channels closely spaced together acquire AEGs that are slowly varying in time, and also share similarities in terms of amplitudes values. Digitizing the signals acquired from every electrode in a multi-channel acquisition system generates a large amount of redundant data. While researchers have attempted to apply various signal specific approaches [1,2,3] such as adaptive resolution ADC, adaptive sampling rate based ADC, a signal specific successive approximation register (SAR) ADC which depends on the difference between the two samples which differ by a few LSBs, a non-linear signal specific SAR ADC, they suffer from being incomplete in the signal data representation and do not account for the redundancies in the spatial domain. In this work, we aim to exploit the redundancies in space based on the gradient or the difference in the amplitudes of the signals recorded between adjacent electrodes. Firstly, signal characteristics of the AEGs are extracted from the available sample data. The difference in amplitude between different channels at a given time instant is calculated and is averaged over many time instants. From a probability density function curve, the mean difference in amplitude and the variance over the entire array is obtained. Based on the values, we define a threshold to activate a high resolution or a low resolution operation mode of the ADC. We aim to develop a signal specific algorithm and circuit architecture using a SAR ADC, that accounts for the changes in the spatial domain to arrive at energy and area efficient multi-channel data acquisition design for AEGs. References: S. O’Driscoll, K. V. Shenoy, and T. H. Meng, “Adaptive resolution ADC array for an implantable neural sensor,” IEEE Transactions on Biomedical Circuits and Systems, vol. 5, no. 2, pp. 120–130, April 2011. E. Rahiminejad, M. Saberi, and R. Lotfi, “A power-efficient signal-specific ADC for sensor-interface applications,” IEEE Transactions on Circuits and Systems II: Express Briefs, vol. 64, no. 9, pp. 1032–1036, Sept 2017. M. Judy, A. M. Sodagar, R. Lotfi, and M. Sawan, “Nonlinear signal-specific ADC for efficient neural recording in brain-machine interfaces,” IEEE Transactions on Biomedical Circuits and Systems, vol. 8, no. 3, pp. 371–381, June 2014. Surgical Process Modelling Strategies: How to Determine Workflow? Maryam Gholinejad, Arjo J. Loeve, Jenny Dankelman Abstract: ABSTRACT INTRODUCTION The vital role of surgeries in healthcare requires a constant attention for improvement. Surgical process modelling is an innovative and rather recently introduced approach for tackling the issues in nowadays complex surgeries. Surgical process modelling allows for evaluating the introduction of new technologies and tools prior to the actual development and is beneficial in optimization of the treatment planning and treatment performance in operating room. This potentially saves considerable cost and effort compared to trial and error development. Therefore, surgical process modelling can potentially aid development of technologies and tools to satisfy the requirements of actual usage experience in the clinical practice. METHODS The concepts associated with surgical process modelling are discussed, aiming to clarify them and to promote their use in future studies. Different modelling strategies are explained and the criteria for opting for the proper modelling strategy are discussed. RESULTS The field of surgical process modelling is complex and still under development. Applying modularity may facilitate and improve the efficiency of surgical process modelling studies and subsequent updates and analyses. If the purpose of a study gives reason to it, the combination of different modelling strategies is advised. Combination of top-down and bottom-up approaches for establishing a surgical process model allows taking advantage of the strengths of both modelling approaches. Similarly, different data acquisition methods could be combined to overcome their individual limitations, achieving a solid, accurate and efficient database on which the development of a workflow or surgical process model can be founded. CONCLUSION Overall, the current review illuminates the importance of surgical process modelling for improving different aspects of treatment procedures and provides an overview of various modelling strategies that can be used to establish surgical process models. Funding: This work is part of the HiPerNav project that received funding from the European Union’s Horizon 2020 Research and Innovation program under grant agreement No 722068. The baby length measuring instrument, The development of a new principle to measure the body length of preterm infants inside neonatal incubators Ronald van Gils, Onno Helder, Harry Broeders, Linda Wauben Abstract: Introduction Accurate length monitoring of preterm infants is, besides bodyweight and head circumference, essential to assess the effect of personalized food intake, and is crucial for optimal neuro-developmental outcome. Currently used measuring instruments (a measuring tape or slide caliper) in a neonatal intensive care unit (NICU) for the weekly body length measurement disturb the infant and are time consuming and unhygienic. Our aim is to develop an accurate and more hygienic length measuring instrument that causes less stress to preterm infants. Methods Through co-design with healthcare professionals of the NICU in the Erasmus MC – Sophia Children’s Hospital, multi-disciplinary students incrementally developed an instrument that measures an infant’s length inside an incubator without physical contact between the instrument and infant. The instrument was used and informally evaluated in the NICU by seven healthcare professionals. Results A length measuring instrument for neonatal incubators was developed that can measure infant’s body length from outside the incubator. The instrument projects two thin light lines parallel to each other into the incubator. One light line is positioned at the head end of the body. The other line is moved towards the feet exterior end of the body. A numeral display shows the distance between the projected lines that corresponds with the body length of the infant. NICU healthcare professionals that used the instrument found it less disturbing for the infant, more hygienic and easy to use compared to the currently used instrument. Conclusions We developed a new principle for measuring a preterm infant’s body length inside an incubator that is more hygienic, less disturbing for the infant and easy to use for healthcare professionals. In following studies, we aim to eliminate the light lines (that can disturb infant) by using two augmented reality images with camera centre lines. Furthermore, we aim to replace 2D with 3D growth monitoring by developing accurate 3D scan technology suited to scan an infant inside an incubator; as a first step, we developed a 3D laser triangulation scanner that generates a 3D height profile of a baby-doll and automatically derives the body length with an accuracy of 5 mm. Respiratory Sinus Arrhythmia in Apnea Patients with Apnea Associated Comorbidities John F Morales Morales, Margot Deviaene, Javier Milagro, Dries Testelmans, Bertien Buyse, Raquel Bailon, Sabine van Huffel, Carolina Varon Abstract: Aim: To improve the phenotyping of apnea patients, it is crucial to consider their associated comorbidities. We hypothesize that the characterization of the cardiorespiratory interactions can be used for this task and for prioriti\-zing the treatment of sleep apnea. Therefore, in this study we quantified the Respiratory Sinus Arrhythmia (RSA) in apnea patients with associated comorbidities. Methods: Electrocardiographic (ECG) and thoracic respiratory signals from Polysomnographic (PSG) studies of 106 patients with different severities of sleep apnea and apnea associated comorbidities, were analyzed (age: 47.3 ± 10.6, Apnea Hypopnea Index (AHI): 37.8 ± 23.8, hyperlipidemia: 52, hypertension: 40, diabetes: 5, heart infarct: 4, stroke: 2). 5 minutes segments, free of apneas/hypopneas during non-REM sleep, were selected from each patient. Next, the R-peaks were detected and the Integral Pulse Frequency Modulation (IPFM) model was used to estimate the Heart Rate Variability (HRV). To quantify the RSA, the -3 dB bandwidth of the Power Spectral Density (PSD) of the respiratory signal was calculated. Afterwards, the power in the same band in the PSD of the HRV was extracted. Finally, this power was normalized with the power between 0.04 Hz and 1 Hz in the PSD of the HRV. Results: A decrease in RSA was observed with age, agreeing with previous studies. However, patients with apnea associated comorbidities and AHI > 35 presented a significant increase of RSA compared to patients without associated comorbidities and AHI < 15 (p<0.05). Additionally, the increases in RSA were unrelated to medication intake. Conclusion: These results suggest an over compensation mechanism in apnea patients with apnea associated comorbidities and higher AHI that might result in an increased vagal tone during non-REM sleep and “normal” respiration. Further research is needed to clearly understand this apparent increased RSA and its potential role for phenotyping apnea patients.