Certain investigations on ischemic Stroke detection from ct brain Images using patching asymmetric Regions and random forest classifier

Abstract

Stroke is the traumatic condition of nerve cells that block down the physical activity of the victims within short span of time. It is the leading reason for the disability and mortality among older population. A timely diagnosis is a crucial step in case of stroke treatment. The modern diagnosis modalities such as Computer-aided tomography and Magnetic resonance imaging have wide application in detecting the stroke lesion. The image processing is a significant process for the segmentation and classification of the medical images. In last two decades, the field of image processing has developed massively and introduced new processing algorithms and machine learning network to process and classify the brain image automatically. Yet, the process has numerous drawbacks that include false positive, network complexity, cost and time effects, processing errors, training process and data constrains. The proposed work focuses on developing a compact system that can reduce the complexity, false positive rate and processing errors which are the serious issues in the existing system. In proposed work the CT brain image with Acute Ischemic Stroke (AIS) lesion is processed and classified using machine learning model. The overall work process of the proposed work is sectioned into two modules. In Module 1, the raw CT brain image is taken from the ISLES 2018 database and processed using Pixel Correlation Histogram analysis to enhance the contrast of the image. In Module 2, the pre-processed image is segmented and classified using Random Forest classifier to segregate normal and AIS stroke lesions newline