BME2022 Paper Submission & Registration
9th Dutch Bio-Medical Engineering Conference

Powered by
© Fyper VOF.
Conference Websites
10:30   Brain
15 mins
Towards an investigational platform for a multimodal neuromodulation approach
Raphael Panskus, Wouter Serdijn, Vasiliki Giagka
Abstract: Over the past decades, neuromodulation has been proven to be an effective treatment for several neurological disorders. Moreover, it continues to be a rapidly evolving field with a wide-ranging potential for biomedical applications. However, efficient and patient-specific targeted neuromodulation remains one of the biggest challenges for implantable devices. Current studies explore the possibility of using multimodal neuromodulation techniques by combining electrical, thermal, optical, ultrasonic, and/or pharmacological modalities to increase the specificity of therapies [1]. Moreover, it is hypothesized that by combining electrical and ultrasonic methods into a hybrid neuromodulation technique, the safety profiles and spatiotemporal resolution could potentially be increased [2]. Low-intensity focused ultrasound has the potential to alter the neural response in a wide range of neuronal targets, with an improved spatial resolution [3], [4]. However, the most effective, reliable, and safe acoustic parameters are currently unknown, especially for the peripheral nervous system, due to the little understanding of the mechanisms that govern this method [5]. In this study, we propose an investigational platform that will allow us to explore a variety of ultrasound parameters for a multimodal neuromodulation approach. The platform integrates a custom-adapted system for stimulation and neural recording, commercially available components for the ultrasound stimulation system, and an experimental control unit with a PC interface. The proposed setup facilitates the evaluation of the tested parameters during experiments on explanted nerve models. Here we will describe potential implementations of such a system and discuss challenges that can be faced during experiments on explanted nerves. This work can be useful to increase our understanding of ultrasound neuromodulation on peripheral nerves and its benefits when integrated into a hybrid platform dedicated to multimodal neuromodulation. [1] S. M. Won, E. Song, J. T. Reeder and J. A. Rogers, “Emerging Modalities and Implantable Technologies for Neuromodulation,” Ultrasound in Medicine & Biology, vol. 45, no. 7, pp. 115-135, 1 April 2020. [2] B. Scheid and S. Chakrabartty, “Feasibility of hybrid ultrasound-electrical nerve stimulation for electroceuticals,” in 2017 IEEE International Symposium on Circuits and Systems, 2017. [3] J. Blackmore, S. Shrivastava, J. Sallet, C. R. Butler and R. O.Cleveland, “Ultrasound Neuromodulation: A Review of Results, Mechanisms and Safety,” Ultrasound in Medicine & Biology, vol. 45, no. 7, pp. 1509-1536, 2019. [4] S. Kawasaki, V. Giagka, M. d. Haas, M. Louwerse, V. Henneken, C. v. Heesch and R. Dekker, “Pressure measurement of geometrically curved ultrasound transducer array for spatially specific stimulation of the vagus nerve,” in IEEE Conference on Neural Engineering (NER), 2019. [5] H. A. S. Kamimura, A. Conti, N. Toschi and E. E. Konofagou, “Ultrasound Neuromodulation: Mechanisms and the Potential of Multimodal Stimulation for Neuronal Function Assessment,” Frontiers in Physics, vol. 26, p. 150, 20 Mai 2020
15 mins
Feasibility of high frame rate Doppler ultrasound during major neonatal non-cardiac surgery
Anna J, Kortenbout, Sophie Costerus, Jurgen de Graaff, Nico de Jong, Jeroen Dudink, Hendrik Vos, Johan Bosch
Abstract: Background and aim: Newborns requiring major surgery have high risk of brain injury, possibly caused by disturbance of cerebral perfusion during surgery. Currently, anaesthesiologists rely on standard monitoring to optimize cerebral blood flow. However, there is no technique that can monitor cerebral perfusion directly. High frame rate (HFR) ultrasound has the potential to provide both high-resolution imaging for vessel visualization and flow quantification in every visible vessel. This study investigates the feasibility of serial HFR Doppler ultrasound during major surgery in neonates and whether quantitative parameter maps provide additional value to conventional monitoring techniques. Methods: Transfontanellar ultrasound measurements at various phases (before induction, after induction of anaesthesia, during surgery and recovery) during high-risk neonatal surgery were performed using a custom HFR mode (>1000 Hz) on a clinical ultrasound system, the Zonare ZS3 with a high frequency linear probe (L20-5). SVD-filtering was applied to the HFR IQ-data to remove tissue signal. A Doppler spectrogram for each blood flow pixel was calculated over time. Spectral peak-velocity detection was done by using an automatic spectral envelope estimation algorithm, which involves estimation of a signal region, a noise region and an envelope cut-off position for every time point. Peak systolic velocity and end diastolic velocity were extracted from the envelope for each blood flow pixel, and based on these values the flowangle-independent Resistivity Index was calculated to generate parameter maps. Results: 98 HFR measurements were performed in ten patients. High-resolution parameter maps of the vascular network of the periventricular cortical zone could be made. The resulting resistivity maps showed both arterial and venous vessels. We observed changes in parameters maps, independently of changes in vital parameters (blood pressure and/or heart rate). Conclusion: In this feasibility study, we showed that simultaneous quantification and visualisation of cortical cerebral blood flow during high-risk neonatal surgery is possible with longitudinal HFR ultrasound and that changes in CBF may occur independently of changes in vital signs. Acknowledgement: This work is part of the MIFFY research programme with project number 15293, which is (partly) financed by NWO. We thank Mindray Innovation Center for the use of the Zonare ultrasound system and the high frame rate mode.
15 mins
The ability of a Real Time Location System (RTLS) and the Electronic Medical Record to monitor hyper-acute workflows and identify bottlenecks. An open, prospective, single center trial in ischemic stroke care.
A (Anisa) Hana, IWF Paulussen, S Chatterjea, J Vervoort, A Leitao, D Ottevanger, RHGM Bisschops, GJ Noordergraaf
Abstract: Main research question: Electronic Medical Record (EMR) data are used as quality indicators for evaluation of hyper-acute clinical pathways. In stroke care, door-to-needle time (DTN) (i.e. IntraVenous Thrombolysis, IVT) should be ≤60 minutes. Validity of this data is uncertain. RTLS technology tracks tags, badges (patients, essential equipment and staff, nurses and technicians; identified at function level in our case) within predefined areas and automatically adds timestamps. The data can be aggregated or individually assessed. Using patients presenting with suspicion of a stroke and monitoring with RTLS supplemented by EMR data, we assessed validity of times, and insights into process inefficiency and badge-wearing acceptability by professionals. Research method: A clinical, single-center prospective, open study using the stroke pathway with enrollment from 1-9-2020 through 31-8-2021. Patients with ischemic stroke and receiving IVT and/or intra-arterial thrombectomy (IAT) were included after informed consent. The RTLS sensors were placed in the Emergency and Radiology Departments, including halls, rooms, and service areas. Patients received tags upon arrival. Interventions into the pathway were introduced into the study (01-07-2021) to validate whether RTLS could visualize changes. Results: High inclusion was achieved (99%, n=125 patients). EMR time point data proved highly inaccurate, RTLS demonstrated high fidelity. For example, maximum difference in DTN time between RTLS and EMR was 26 minutes. The overall compliance of wearing a badge by a required professional for that step in the workflow was 81 ± 16%. Compliance (acceptability) was variable: nurses 96%, Neurology residents 64%, and Radiology technicians 85%. Patient tag signals were not visible for RTLS (i.c. temporarily under covering) at least once in 74% of cases, typically during CT-scans. We were able to recognize and suggest adaptations in bottlenecks in the stroke pathway resulting in 25% time saving through a) using real entry-into-workflow time; b) improving content with IVT and IAT box, by bundling materials thus allowing the nurse to stay in the admission rooms; and c) creating a dedicated Anesthesia cart including 24/7 stand-by Anesthesia care station. Conclusions: Data from RTLS, with EMR data offers a detailed, independent insight into hyper-acute care and discerns inefficient constituents. Professionals are ambivalent about badge-wearing.
15 mins
Early sepsis detection on the ward using deep learning
Sebastiaan Oei, Ruud van Sloun, Myrthe van der Ven, Hendrikus Korsten, Massimo Mischi
Abstract: Background Sepsis is one of the leading causes of death in the hospital. Several warning scores have been developed to categorize patients’ degrees of illness, with the purpose of recognizing sepsis onset at an early stage and consequently reducing time before starting treatment. The most accurate classification method, known as the SOFA score, is developed for use in the intensive care unit (ICU). Objective Sepsis is not exclusively developing in the ICU and may occur in any hospitalized patient. Therefore, a reliable method for sepsis recognition using routinely gathered information from the general ward is of major importance. Methods Recently, the use of computational methods has been proposed for early sepsis prediction. Multiple sepsis classifiers have been devised using machine learning methods. We validated a linear classification model and improved upon it using a deep neural network trained on data from the MIMIC-III database. Results The reference model approach yielded an Area Under Precision-Recall Curve (AUPRC) of 0.45 for a 3-h prediction time. Our newly constructed deep neural network outperformed the reference model, reaching an AUPRC of 0.49 for a 3-h prediction time. These results demonstrate to be robust for multiple prediction times, utilizing different datasets and multiple sepsis criteria. Conclusions Our results are comparable to a high-resolution model, yet using only 8 simple and commonly performed measurements, instead of the complex set of dozens of measurements leveraged by other authors. Moreover, our method proves to adapt to different datasets and sepsis criteria, yielding a high accuracy. Therefore, sepsis onset prediction may also be viable in less monitored environments in the hospital, such as the general ward and the emergency room.
15 mins
Development of a classifier to identify patients with major or persistent depression disorder using smartphone and wearable data
Ahnjili ZhuParris, Ghobad Maleki, Robert-Jan Doll, Geert Jan Groeneveld, Gabriel Jacobs
Abstract: Currently, the efficacy of therapeutic interventions for mood disorders is predominately evaluated using in-clinic questionnaire-based assessments. This approach is limited in its ability to capture symptom-related changes in real-time and outside the clinic. Real-world data related to physical activity, sleep, and social behaviour could potentially provide additional clinically relevant insights into factors influencing mood disorder symptom severity. Furthermore, these data may be useful for assessing treatment effects. The objective of this study is to develop a depression classification model using data collected from smartphones and wearables. Data used for modelling was collected during a 3-week non-interventional pilot study including 30 Major Depressive Disorder (MDD) or Persistent Depressive Disorder (PDD) and 29 control subjects. Stable antidepressant use (minimum 4 weeks) was allowed for the depressed subjects. The MORE™ platform sourced data from the subjects’ smartphone and three biometric sensors: a smartwatch, a smart-scale, and a smart blood pressure monitor. To create the optimal classifier, we compared 3 types of feature selection methods (No feature selection prior to modelling, Variance Thresholding and Variance Inflation Factor) and 3 Machine Learning classifiers (Logistic Regression, Random Forest Classifier, and Gradient Boost Classifier). We developed 56 features including acceleration magnitudes, (smartphone) app usage, call logs, human voice detection, and location data. The biometric sensors provided features relating to sleep, heart rate, steps, weight, fat and muscle percentage, and blood pressure. The optimal model was a random forest classifier validated using a 5-fold cross validation strategy (no feature selection was applied). The model classified achieved a 69% accuracy, 69% sensitivity, and 70% specificity. In order of descending importance, ambient human voice detection, call duration, and number of interactions with social apps were the most salient features that distinguished depressed from non-depressed subjects. We demonstrated the ability to classify depressed and non-depressed control subjects using smartphones and wearables. More specifically, features that captured smartphone use, social behaviours and mobility were the most informative for this classification. This approach demonstrated the potential to continuously monitor behavioural biomarkers for future drug trials in depression.
15 mins
Influence of anisotropic electrical conductivity in white matter tissue on the EEG source reconstruction accuracy
Stefan Dukic, Boudewijn T.H.M. Sleutjes, Leonard H. van den Berg
Abstract: Source reconstruction of cerebral activity using EEG/MEG data is becoming an established tool in neuroscientific research. Recently, using source analysis on EEG data we have shown multiple networks that are affected in amyotrophic lateral sclerosis (ALS),1 some of which we have not observed using sensor-space analysis.2 This approach, however, requires a volume conductor model of the human head that closely mimics the electromagnetic properties of the investigated subject as accurately as possible. These models often assume isotropic conductivity tensors across different tissues, which is especially known to be incorrect for the skull and white matter. Here, we investigate the influence of white matter anisotropy derived from diffusion tensor imaging (DTI) data on the dipole estimation error. To construct an anisotropic head model, a T1-weighted and a DTI dataset were acquired from a healthy volunteer (female, 60 years). We segmented the MRI scan into six compartments (i.e. grey and white matter, cerebrospinal fluid, skull, scalp and air cavities). For the finite element mesh generation hexahedral elements were used. Based on this geometry, two models were made: 1) a simple model with isotropic conductivity tensors assigned to elements belonging to the same tissue type and 2) an advanced model with anisotropic conductivity tensors assigned to elements belonging to the white matter. Anisotropic conductivity tensors were estimated using DTI-derived diffusion tensors and the linear relationship between the two estimates.3 The dipole location error is determined by applying dipole fitting 125 times on each model using simulated EEG data (15 Hz sinusoid originating from the right putamen). Using Euclidean distance, the localisation error was on average 6.28 mm (range: 0.13 - 44.43 mm) for the simple model and 6.07 mm (range: 0.09 - 32.71 mm) for the advanced model indicating slight advantages for the advanced model. This study shows evidence of the importance of white matter anisotropy modelling in healthy individuals. Accounting for white matter anisotropy is likely to have an even greater impact in diseases that affect white matter, like ALS. Further analyses which use more repetitions (>125) and that assess other dipole locations (beyond the putamen) are warrant. [1] Dukic, S. et al. Patterned functional network disruption in amyotrophic lateral sclerosis. Hum. Brain Mapp 40, 4827–4842 (2019). [2] Nasseroleslami, B. et al. Characteristic increases in EEG connectivity correlate with changes of structural MRI in amyotrophic lateral sclerosis. Cereb. Cortex 29, 27–41 (2019). [3] Rullmann, M. et al. EEG source analysis of epileptiform activity using a 1 mm anisotropic hexahedra finite element head model. Neuroimage 44, 399–410 (2009).

end %-->