Institute of Human Movement Sciences and SportOpen OpportunitiesThis master’s thesis is dedicated to developing an advanced nutrition tracking system for hospitals, integrating QR-code recognition and structured light camera technology. The focus is to significantly enhance the precision of food volume measurements and patient meal tracking with machine learning, thereby improving nutritional monitoring accuracy. - Computer Vision, Medical and Health Sciences, Software Engineering
- Master Thesis
| Neurofeedback (NF) is a promising approach for training healthy participants and patients to modulate their motor-related neural activity even in the absence of overt motor output. Motor imagery (MI)-based training, i.e., participants mentally simulate movements, also has beneficial effects on the restoration of impaired motor function. Transcranial magnetic stimulation (TMS) is a non-invasive, low-risk method that is routinely used for psychological or neuroscientific research in human participants. In comparison to electroencephalography, TMS-based NF has great potential to distinguish fine-grained MI tasks such as different hand actions. This is important because daily life activities require complex coordination of hand muscles. A hand function training is critical for individuals with impaired hand function. Our group has developed a new protocol that uses TMS to detect MI-induced motor activity patterns in the primary motor cortex. Here we will use TMS over the primary motor cortex of participants to measure motor evoked potentials (MEPs) in finger muscles during either motor execution or motor imagery of different hand actions, namely holding a bottle, turning a key and opening the hand. Based on the MEPs we provide participants visual feedback. We aim to further develop and validate an online, adaptive classification algorithm that decodes imagined hand actions in healthy volunteers from TMS-evoked MEPs and potentially apply this to stroke survivors.
Our group has recently completed the pilot data acquisition investigating the performance of an adaptive classification algorithm for decoding imaged hand actions during TMS-based NF training. We will continue with data collection and include brain MRI scans to further develop and validate this novel TMS-based NF training protocol. - Motor Control
- Master Thesis
| This thesis aims to enhance food volume estimation in healthcare settings, a critical factor in patient nutrition management. The project involves developing a machine learning system capable of synthesizing 3D food images. The enhanced accuracy of volume estimations will be achieved through an expanded training set enriched with real-world food data from hospitals. This research, conducted by an ETH team, is poised to significantly impact patient care in facilities, particularly in addressing malnutrition. The initiative is not just academic but also a stepping stone towards a potential startup venture, emphasizing the project's practical applicability and future growth potential. - Computer Vision, Image Processing, Neural Networks, Genetic Alogrithms and Fuzzy Logic
- Master Thesis
|
|