A CNN-Model to Classify Low-Grade and High-Grade Glioma From MRI Images

Glioma is the most occurring brain tumor in world. Its grade (level of severity) identification, crucial its treatment planning, demanding a clinical environment. Computer-aided methods have been experimented with to identify glioma, out which deep learning-based methods, due their auto features engineering, good impact terms achieved outcomes. In this study, convolutional neural networks (CNNs) explored and utilized for classification glioma grading, example, low (grade I-II) high III-IV). A CNN-based model, light-weighted layers, size, learnable parameters, has proposed. Experimental tests were carried on benchmarked publicly available datasets, Brats-2017, Brats-2018, & Brats-2019. locally developed dataset from Bahawal Victoria Hospital, Bahawalpur, Pakistan, also employed experimentation research cross-validate Additionally, experiments compare effectiveness proposed results compared state-of-the-art pertained CNN models, i.e., resnet18, squeezenet, alexnet. The model maximum standard evaluation measures dataset, accuracy, specificity, sensitivity at 97.85%, 98.88%, 99.88%, respectively. Similarly, these 98.89%, 99.28%, 99.77% best recent related studies.