ABSTRACT
Automated identification of brain lesions are always challenging, as there would be a difference in the composition of grey and white matter, which would reflect different intensity levels in the imaging. However, a prediction can be drawn based on the variation with respect to the contralateral side of one's own image, and based on the knowledge gained from previous studies. The most common and most reported brain lesions are from strokes. Among the two types of stroke, the ischemic stroke is most challenging to identify, as there isn't be much variation in the grey levels between the affected and nonaffected regions. This chapter will provide an insight into various steps, which can be adapted while processing the images of both ischemic and hemorrhagic strokes to segment the region of interest. Although these steps are determined by various factors, this method will inform researchers on the preprocessing required and how to identify a suitable feature extraction method for further segmentation. A case study is presented and expanded upon with examples for better understanding.
