Semi-invasive 3D Mapping System for Cardiac Arrhythmias
Eyetracking for Sports
Mess- und Informationssystem für Monitoring in der Fussball-Ausbildung
Dynamic Contour Tonometer Lens
Arc Fault Detection in PV Systems
Body Sensor Network
Parallelization of OCT Hardware Algorithms
Optical Coherence Tomography Engine
The E2corder (Esophageal ECG Recorder)
Real-Time Dietary Assesement and Physical Activity Analysis in Dietry Counseling
Motion Analysis and Signal Processing in Race Bicycles
Smart DAQ Time-of-Flight for Mass Spectrometry
GECKO: General Purpose System-on-Chip HW/SW Co-Design Environment

Semi-invasive 3D Mapping System for Cardiac Arrhythmias

esophagus ECG catheter design

Marcel Jacomet, Josef Götte, Thomas Niederhauser, Lukas Bösch, Andreas Moser, Luca Fleischmann, Reto Wildhaber, Dominik Robellaz, Dominik Brügger, Radhakrishna Rakesh Vasireddy, Christof Baeriswyl, Frédéric Waldmann, Emily Kathryn Thompson, Gerhard Frédéric Kuert, Marco Dubach — May 2016 - May 2019

Partner Insel Hospital: Hildegard Tanner, Andreas Häberlin, Romy Sweda

Heart rhythm disturbances (arrhythmias) are common and may pose an imminent threat to patients. Some arrhythmias are responsible for a highly increased risk of ischemic strokes leading to a higher morbidity and mortality. Therefore, early detection and prevention of adverse arrhythmia-related events are crucial. In case where an interventional treatment of arrhythmias is indicated, the planning of the intervention strongly relies on the type of arrhythmia diagnosed based on standard 12-channel body surface electrocardiograms (ECGs). The diagnostic accuracy of this standard approach is however limited. Our approach is to complement these 12-channel surface ECGs by ECGs captured additionally from inside of the esophagus. The signal quality of these esophageal ECGs is superior to the surface ECGs due to the proximity of the measuring electrodes to the origin of the electrical signals from the heart’s muscle, the myocardium; additionally, we devise specially constructed catheters that allow a 3-dimensional description oft the heart depolarization process. We expect that our new approach outperforms the diagnostic accuracy of the 12-channel ECGs and renders itself suitable to bedside decision making by providing more detailed information on arrhythmias. The research procedure reveals the tight link and strong interdependency of the five steps: the engineering side develops catheters, corresponding measurement techniques, signal processing and hardware algorithms; the medical side uses these tools in their clinical work to produce measurement data, to validate the computational results, and to gain deeper insights for improving diagnosis and bedside decision making based on more detailed information on arrhythmias.

Kampf der Herzrhythmusstörungen - Face2Face Meeting 22 Februar 2018

Eyetracking for Sports

Sport Outdoor Brille mit 5 integrierten FPGAs für high-speed Eye-Tracking Hardware-Algorithmen

Marcel Jacomet, Reto Pablo Meier, Damian Weber, Ties Jan Kluter — May 2016 - December 2017

Hochschul-Partner Universität Bern: Enst-Joachim Hossner, Ralf Kredal, Carlo Prelz

Die Analyse des Blickverhaltens rückt in den letzten Jahren zunehmend ins sportwissenschaftliche Interesse. Bis jetzt verhindert jedoch das Fehlen einer geeigneten Messtechnik den breiten Einsatz in der Sportpraxis. Daher entwickeln wir zusammen mit dem Institut für Sportwissenschaft ISPW der Universität Bern einen sporttauglichen Eye-Tracker. Höchste Anforderungen stellt dabei vor allem der hohe Miniaturisierungsgrad, denn in der Brille sind für die Auswertung der Front-Kamera und der 4 Augen-Kameras total 5 FPGAs mit entsprechenden Hardware-Algorithmen für das Eye-Tracking implementiert.

EyeTracker

Mess- und Informationssystem für Monitoring in der Fussball-Ausbildung

Sensoren und Algorithmen zur Extraktion von Schlüsselinidkatoren im Fussball

Benjamin Habegger, Marcel Jacomet, Josef Götte, Damian Weber — May 2016 - April 2018

Industrie-Partner Aximao GmbH: Michael Gasser

Hochschul-Partner Baspo/EHSM: Urs Mäder, Martin Rumo

In diesem Forschungsprojekt extrahieren wir kritische Erfolgsfaktoren im Fussball über objektive Schlüsselindikatoren aus Sensordaten. Wir wollen diese Zielsetzung mit dem Einsatz von tragbaren Mini-Sensorknoten (Axiomote vom BFH-HuCE start-up Aximao GmbH) in einem Netzwerk und entsprechenden Datenfusions-Algorithmen erreichen. Die Kombination aus Inertialsensorik und Positionsmessung erlaubt erstmals die Verbindung von Spieler-Aktionen und Attributen mit den Spieler-Positionen, und kann Fragen beantworten wie z.B. "Von wo aus wird wie stark geschossen". Das resultierende Mess- und Informationssystem hat das Ziel, die Leistungsoptimierung im Training, sowohl des Einzelspielers als auch der Mannschaft, zu verbessern. Die wissenschaftliche Innovation ist die Leistungsmodellierung nach kritischen Erfolgsfaktoren (Critical Success Factors, CSF) und den zugrunde liegenden Schlüsselindikatoren (Key Performance Indicators, KPI). Aus den Sensordaten werden Muster erkannt und daraus Spieler-Aktionen bestimmt; aus den Spieler-Aktionen werden die definierten Schlüsselindikatoren berechnet; daraus bestimmen sich kritische Erfolgsfaktoren, Leistungskonstrukte und schlussendlich die Spielleistung. Diese Modellierung erlaubt aus den objektiven, technischen Messungen von Position und Inertialvariablen die relevanten Informationen zu extrahieren.


Dynamic Contour Tonometer Lens

Low Power, Wirless Pressure Sensing and Recording Electronics

Marcel Jacomet, Josef Götte, Damian Weber — May 2012 - August 2013

Glaucoma is an eye disease that is the second most reason for blindness. An elevation of the intraocular pressure (IOP) is the highest risk factor for this disease. Goal of the present research project is to develop of a pressure measuring system for continuous measuring the IOP during day and night.

Dynamic Contour Tonometer Lens

Arc Fault Detection in PV Systems

Arc Fault Detection in Photovoltaic Applications

Marcel Jacomet, Josef Götte, Benjamin Grichting, Marcel Jost, Daniel Gfeller — January 2012 - December 2012

Arc Faults in PV systems can cause great damage in the system itself or the close surrounding. The project goal is to develop an algorithm that detects such an Arc Fault with a high reliability. The algorithm must be implemented as simple as possible in consideration of computational needs, due to an implementation in existing inverters. In collaboration with the Photovoltaic-Laboratory we created a database of several hundreds of measured Arc Faults in different situations. The database contains a rich variety of different measured signal properties, for instance variation of noise or spectral properties and its variations to name just a few. Based on the analyzed measured data an algorithm has been developed that enables the detection of Arc Faults and avoids a faulty activation of switching operations. Finally, the algorithm was implemented in a autonomous system in collaboration with the research group IMC-Embedded-Systems. The implementation provides us with further in field tests.


Body Sensor Network

Sensor nodes are integrated into a local positioning system (LPM) for real-time sport games analysis

Benjamin Habegger, Marcel Jacomet, Josef Götte, Michael Gasser — November 2011 - June 2014

The physical activity of human beings in daily live, at work, in military service and during sports shall be logged and analyzed. The key features of the physical activity recording (and tracking) network (PARTnet) device is its small packaging dimensions and low power consumption of its node devices. The final goal is to integrate the device into a thin and wearable bracelet or upper arm belt of high wear comfort. The low power consumption is the basis for activity logging during several weeks or even months. We develop a tiny, wireless body-sensor network, which is able to capture and log various data from different sensors. The central device houses a 3D accelerator sensor and features a large memory for data logging. In addition to these accelerator data, the device logs in a time-synchronous way sensor data from other wireless devices as well. Presently, we have additional 3D accelerator sensor nodes distributed over the human body and a commercially available heart-rate sensor belt. The various kind of sensors are used to extract and to monitor physical activity and to draw conclusions on physical stress. In a final step, the body sensors are integrated into a commercially available local positioning measuring (LPM) system, developed by Abatec Inc, Austria for analyzing sport games on-line.

PARTwear Platform
Body Sensor Network

Parallelization of OCT Hardware Algorithms

OCT implementation on small credit card size FPGA board

Andreas Habegger, Benjamin Habegger, Marcel Jacomet, Josef Götte, Reto Pablo Meier, Pascal Kesselring — October 2011 - March 2013

Optical Coherence Tomography (OCT) is an interferometric optical signal acquisition and processing method used in application with micrometer resolutions. A relative recent implementation of OCT, the so called frequency-domain OCT, provides advantages in signal-to-noise ratio. Providing high-speed signal processing hardware, volumetric images can be captured and processed with real-time constraints. OCT can be used in diverse applications, art conservation, diagnostic medicine like ophthalmology and others, where images with high resolution are needed. The goal of this project is to investigate in hardware parallelization techniques for frequency domain OCT algorithms such that very-high speed processsing can be achieved with low cost FPGA fabrics, running on the GECK4 co-design plattform. The goal is to achieve a continuous processing speed of at least 200 MS/s on a low-end Xilinx Spartan family FPGA. A compact parallelization architecture has to be developed and its performance and complexity verified with Simulink.

Low-Cost OCT Hardware Algorithms

Optical Coherence Tomography Engine

High-Speed Hardware Algorithm Design for Signal Processing

Marcel Jacomet, Josef Götte, Vinzenz Bandi — June 2010 - May 2012

The goal of the project is to develop a complete miniaturized OCT system, based on the newest generation of rapid swept source from Exalos. Data processing algorithms are implemented in hardware to provide fully processed images in real time. This OCT engine finds applications in ophthalmology, in dentistry, in general medical diagnostics, and in quality assessment in industrial production. Our tiny OCT engine consists of the miniaturized swept source, the interferometer, the delivering optics, the detection electronics, and the signal processing unit. We implement the computation-power intensive signal processing for the frequency-domain OCT as a real-time hardware algorithm on an FPGA. The OCT engine achieves an overall and steady A-scan rate of 110 kS/s at 2048 pixels per A-scan. The scientific objectives include the realization of a novel optical measuring principle to achieve equidistant sampling in frequency space, and the implementation of an algorithm to handle the system-inherent mirror ambiguity. Moreover, we research in highly parallel and thus high-speed OCT hardware-algorithm design for pre- and post-processing of A/B-scans in real time, and we investigate novel data compression methods to solve the problem of the huge data-transfer rate with standard PC interfaces.

Low-Cost OCT Hardware Algorithms

The E2corder (Esophageal ECG Recorder)

An esophageal implantable device to record the heart rhythm during 30 days

Andreas Habegger, Marcel Jacomet, Josef Götte, Thanks Marisa, Thomas Niederhauser, Lukas Bösch, Michael Nydegger, Sandro Burn, Caspar Trittibach, Reto Wildhaber, David Metzger, Andreas Moser, Nicolas Balmer, Dominik Robellaz — January 2010 - December 2018

Cardiac arrhythmia are often accompanied by palpitation which a patient may not interpret as a rhythm disorder. To detect such arrhythmia cardiologists need a long-term capturing device for ECG signals. The project develops a recorder that measures ECG signals from inside of the esophagus close to the heart. As compared to existing Holter devices, our ECG recorder is able to capture long-term signals inherently in a much higher quality. First we develop an E2coder system consisting of a sensor-catheter and a behind-the-ear electronics platform. We investigate the novel ECG signal-types that result from esophageal recording and research on their clinical interpretation as well as their signal compressibility. Second, with the acquired novel knowledge we built an ASIC that allows the signal processing to reside in the esophagus as well.

Esophageal ECG Recorder

Real-Time Dietary Assesement and Physical Activity Analysis in Dietry Counseling

A smart-phone application

Marcel Jacomet, Damian Weber — January 2010 - January 2012

Many overweight/obese adults in the EU need reliable dietary advice. Attractive tools help to log, to analyze and to display dietary intake and physical activity data over time to provide personalized support and enhance compliance in dietary counseling. Our objectives are to develop and test a smartphone application (smartERB) for integrated use in face-to-face dietary counseling of overweight/obese clients. Synchronize real-time assessment of dietary intake and automated capturing and analysis of physical activity, and visualize the resulting energy data as a motivational component in dietary counseling.

Smart Phone App

Motion Analysis and Signal Processing in Race Bicycles

Low power wireless sensor network

Marcel Jacomet, Josef Götte, Michael Gasser — October 2009 - November 2011

We perform dynamic motion-analyses of high performance road-racing bicycles featuring carbon frames. We achieve our goals by capturing measurements from various wireless sensor nodes and applying sophisticated signal analyses to the collected data. The project explores the relation between statically and dynamically determined frame stiffness; it also explores the impact of stiffness on traction and on rider's comfort. We capture different variables describing the racing bicycle and its rider to get an as complete as possible picture: 3D accelerations at frame positions, pedal frequency, pedal force, GPS position, and heart rate. These variables are captured at various positions of the bike and the rider. For that purpose, we develop tiny autonomous sensor nodes which communicate with each other. The data analysis will deliver design hints to optimize the frame construction.

Motion Analysis in Race Bikes

Smart DAQ Time-of-Flight for Mass Spectrometry

High-Speed Data Acquisition and Signal processing Hardware Algorithms

Marcel Jacomet, Josef Götte, Roman Held — May 2009 - November 2011

We realized a real-time, high-speed data-acquisition and data-processing system for continuous mode time-of-flight mass spectrometers. To achieve data acquisition rates of 2 x 1.5 giga samples per second (GS/s), needed for the considered class of mass spectrometers, we implement the system as an FPGA based hardware algorithm. We must solve two most challenging problems: First, the high-speed acquisition produces an enormous amount of data that we handle by on-the-fly data compression/ uncompression to circumvent the memory-bandwidth restrictions. Second, the need for continuous acquisition of mass spectra and event-triggering ask for powerful hardware algorithms that allow to measure long signals that are composed of ultra-short signal pulses due to single aerosol- or nano particles.

High-Speed Hardware Algorithms in Mass Spectrometry

GECKO: General Purpose System-on-Chip HW/SW Co-Design Environment

An educational robot and SOC research platform

Andreas Habegger, Benjamin Habegger, Marcel Jacomet — January 2009 - December 2012

The latest member of the Gecko family, the Gecko4main module, is an experimental platform that offers the necessary computing power to speed intensive real-time algorithms and the necessary flexibility for control intensive software tasks. Adding extension boards equipped with sensors, motors and a mechanical housing suits the Gecko4main module for all manner of industrial, research and educational projects. Currently a driver for Mathworks Inc Simulink tool is under development, allowing to directly download Simulink designs as either hardware or software implmentations on the Gecko4main board.

Xcellence: General-Purpose SoC Plattform Handles Hardware/Software Co-Design