Predetection of Stroke by Using Heuristics and Artificial Neural Networks

Abstract

The strokes are an important cause of death for all the people, not only for the aged population. Sooner a stroke is discovered better chances are for the patient to minimize the damage or even to survive it. The complexity of strokes reveals clearly the importance of early stroke predetection which are not only helping the doctors but they could literally save lives. Algorithms for predetection of stroke are diverse, however they are little explored. This thesis is mainly centered on predetection of stroke, based on the inversion of T waves in electrocardiograms. Two models were proposed in this thesis to help providing efficient predetection of stroke for people suffering of myocardium diseases and myocardial ischemia. The algorithms were tested on data from four electrocardiograms given by a library and five electrocardiograms from five different patients. Filters for noisy signals are also provided in this thesis. These algorithms can be used as a tool by nurses and doctors but they do not represent a fully automated detection of stroke.