Brain Tumor Segmentation and Evaluation Empowered with Deep Learning

A critical stage in the diagnosis and planning of brain tumor treatment is the segmentation of the tumor. Deep learning approaches have lately demonstrated considerable potential for properly segmenting brain cancers from MRI and CT data. In this research study, a deep learning-based comparative strategy is used for segmenting and evaluating brain tumors using U-Net, Residual Network, and Multi ResNet architectures are compared. The deep neural networking model used in the proposed method is trained using a sizable dataset of MRI brain image scans, enabling it to distinguish between tumor-affected regions and normal, healthy brain tissue. The multi-ResNet architecture is utilised to accurately separate the tumor regions and capture the high-level features of the brain images. Proposed experimental comparison demonstrates that the multi residual network achieves high accuracy and outperforms traditional methods on benchmark datasets. Among the three algorithms performed in this study, Multi ResNet marked dice coefficient of 0.89, Precision of 0.91 and Recall of 0.87.