The aim of this project is the PhysioNet’s and Computers in Cardiology challenge of 2003, specifically the building of a model of ST Segments, based on component analysis, and the creation of a classifier that can categorize these segments to ischaemic or non-ischaemic. Two techniques were used to visualize the data, plots of Principal Components and Neuroscale, with various datasets. However, these techniques performed poorly because they did not separate the two classes in two dimensions. These datasets were also used for classification. Using only the extracted Principal Components the results were poor when compared with the other entries of the challenge. Adding ΔST and ΔT into our dataset the results improved remarkably. The best classifier created with that dataset had accuracy of 89.1%. Finally, using Automatic Relevance Determination method we conclude that ΔT is the most significant variable in classifying ischaemia.
| Date of Award | 2006 |
|---|
| Original language | English |
|---|
| Awarding Institution | |
|---|
- ECG analysis
- electroencephalography (EEG)
- information engineering
ECG Analysis
Smirnakis, M. (Author). 2006
Student thesis: Master's Thesis › Master of Science (by Research)