Survival prediction in glioblastoma brain tumor using segmentation and Detection with advance computational techniques

Abstract

In recent times, a wealth of evidence has emerged, indicating a notable increase in braintumor cases, solidifying its status as the 10th most prevalent type of tumor, affectingboth children and adults. Glioma tumors, assessed pathologically, are divided into theformidable glioblastoma (GBM/HGG) and the less aggressive lower grade glioma(LGG). Glioblastoma, among various brain tumors, stands out as the most lethal andaggressive. Within gliomas, diverse histological subfields include peritumoral edema,a necrotic core, and enhancing/non-enhancing tumor cores. Radiology, specifically magnetic resonance imaging (MRI), plays a vital role in unraveling the phenotypicintricacies and intrinsic heterogeneity of gliomas. Utilizing multimodal MRI scans,such as T1-weighted, contrast-enhanced T1-weighted (T1GD), T2-weighted, andfluidattenuation inversion recovery (FLAIR) images, provides a holistic understanding ofdifferent glioma subfields. The need for precise predictions in overall survival,diagnosis, and treatment planning for glioma patients is met through automated newlinealgorithms embedded in a brain tumor segmentation and detection framework. Thesealgorithms leverage fragmented tumor subfields and radiometric characteristics frommultimodal MRI scans. The thesis introduces a model framework encompassing tumor newline newline