Comparative Analysis of Deep Learning Approach for Detection and Segmentation of Brain Tumor

A brain tumor is one of the more aggressive diseases that affect both children and adults. The survival rate for patients with brain cancer or CNS tumors is approximately 34% for men and 36% for women. If not detected promptly and accurately, brain tumors can result in death. To extend the patient's life expectancy, appropriate treatment, planning, and precise diagnostics are required. Magnetic Resonance Imaging (MRI) is the best method for finding brain tumors. However, MRI calls for a skilled neurosurgeon to conduct an in-depth analysis. In the field of Computer Vision (CV), the emergence of DL offers cutting-edge solutions to image processing issues. This study used the Convolution Neural Network (CNN) algorithm and five Deep Learning (DL) models to use Transfer Learning (TL) to classify brain tumors. The accuracy of the ResNet50 and DenseNet121 models was 93.23%. However, the VGG16 model performed the best, with an accuracy of 97.08%, followed by MobileNetV2’s accuracy of 97.02, and DenseNet121 had the lowest accuracy, with 92.86%.