master's thesis

Transformed-Based Myoelectric Decoding for Continuous Control of Prosthetic Fingers

This project was my master’s thesis for the Master of Science by Research degree part of the UKRI CDT in Biomedical AI at the University of Edinburgh. I worked on a transformer model with multiple outputs, classiyfing electromyographic (EMG) recordings at the wrist into finger movements:

The idea was that if this can be done on sufficiently short windows of EMG, we could attain continuous and independent control of prosthetic fingers. With a substantial performance increase compared to previous attempts in the lab I showed that this paradigm is feasible and further research should be fruitful.

If you are interested, take a look at the thesis here.
Additionally, I published a summary of the results on Weights & Biases.

Supervisors: Prof. Kia Nazarpour & Dr. Chenfei Ma.