Classification of ischemic stroke subtypes using artificial intelligence methods

Millions of people worldwide are known to have a stroke every year. Stroke can also can also cause people to lose their lives and experience physical problems. Since stroke is a type of disease that occurs suddenly,rapid intervention is required. Stroke lesion detection is performed by neurology or radiology specialists using biomedical imaging methods. The high number of image acquisition in hospitals and the low number of specialist physicians and manuel segmentation is a time-consuming process and may prolong the process of stroke diagnosis. Using Diffusion MR Imaging(MRI) image technique, we proposed a model for shortening the detection process of ischemic stroke with a model we call Pyramidal and Fixed Dimensional Patch Based Feature Engineering Model(PFP-FE). For the accuracy of our proposed PFP-FE model for ischemic stroke detection and classification, we created a new dataset containing 3673 Diffusion MRIs. The accuracy rate of the proposed model was calculated as 87.56%