Automatic detection of myoclonus bursts: A follow-up study

Kwant, Joy (2024) Automatic detection of myoclonus bursts: A follow-up study. Master's Internship Report, Computing Science.

Preview	Text mCS2024KwantJ.pdf Download (5MB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (129kB)

Abstract

Myoclonus refers to brief, shock-like involuntary muscle contractions that can vary in intensity and frequency. Electromyography (EMG) signals measure muscle activity and can record myoclonus bursts in patients with this movement disorder. Currently, clinicians manually detect myoclonus bursts, a process that requires expert analysis and is time-consuming due to the significant amount of EMG data. Machine learning offers potential solution to this problem. This research is a follow-up to the study “Automatic detection of myoclonus bursts”, which made the first step in myoclonus burst detection using machine learning. The focus of this study is to potentially improve burst detection by using hybrid models that combine Convolutional Neural Networks with Long Short-Term Memory (CNN+LSTM) and Gated Recurrent Units (CNN+GRU). Both models demonstrated the ability to identify bursts that corresponded to the annotated bursts in the EMG data of myoclonus patients Additionally, the models predicted bursts in the EMG data of healthy controls, which only contains non-bursts data. Although the models have shown improvements and potential in detecting myoclonus bursts, further refinements are needed to increase their accuracy and reliability.

Item Type:	Thesis (Master's Internship Report)
Supervisor name:	Biehl, M. and Brandhof, E.L. van den and Dalenberg, J.R.
Degree programme:	Computing Science
Thesis type:	Master's Internship Report
Language:	English
Date Deposited:	30 Aug 2024 13:17
Last Modified:	30 Aug 2024 13:17
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/34130

Actions (login required)

View Item