A Systematic Review of Current Advances in Ischemic Stroke Detection and Segmentation

Abstract

Ischemic stroke is now one of the vital factors for disability and mortality that globally affects millions of individuals each year in accordance with the World Health Organization (WHO) contrast to hemorrhagic stroke. Treatment for an ischemic stroke as soon as possible can assist to limit prolonged damage and even decreases the risk of mortality. The diagnosis is based on a neurologist's visual observation, which may differ from one to another. On the other hand, Manual segmentation is a tedious and instinctive procedure that has a conspicuous impact on Acute ischemic stroke encountered patient’s prognosis. Numerous automated computer Aided Diagnosis (CAD) systems dependent on many statistical learning algorithms of machine learning (ML) and multi-neural network architecture of deep learning (DL) were considered to reduce the complexity of prediction and lesion segmentation in ischemic stroke and also lower the time required for the manual procedure. This paper contemplates the Imaging modalities, Pre-processing techniques, and segmentation algorithms of ischemic stroke, as well as their performance based on comparing different evaluation parameters and their disadvantages. It highlights the current needs, preferred modality, and possible research ideas in the stroke sector. Keyword : Brain MRI; Deep Learning; Ischemia; Machine Learning; Pre-Processing;