Electromyography can be used as a human machine interface in which a person could control a computer or device with the electrical signals that cause muscle movement. This research used EMG data from the SEEDS data base to explore what features should be used by machine learning algorithms to accurately classify EMG data into which motion a subject is preforming. We extracted features from the EMG data and then ran three different feature selection algorithms to find which features were the most useful in classification. In the end, we found eight features that our various selection algorithms selected the most and concluded that those features would be good starting place when trying to classify EMG data.
The team's goal was to design a hardware system that combines an ultrasound medical imaging device with additional imaging modalities. This new system will simplify the current procedure for more effective diagnosis, decrease preparation and cleaning times, and improve the experience for the patients.