MRI Brain tumor segmentation and classification with improved U-Net model

Brain tumors are among the deadliest diseases in the world. Early diagnosis thereby improves the patient's prospects and likelihood of recovery. It takes a lot of work to separate cancer from other brain abnormalities from MRI images. Various approaches have been devised to predict and divide the tumor. Choosing the finest feature extractor, the long run-time need, and the necessity for expert support are just a few of the challenges they encounter. Improved U-Net and a White Shark aided Beluga Whale Optimization based DCNN are introduced to segment and classify brain tumors (WSBWO based DCNN) into four stages in order to address such concerns. This work uses an improved U-Net based image segmentation model, and the first step of MF based preprocessing uses the input image. The third stage is feature extraction, when statistical features such as I-GBP and MTH are extracted. Subsequently, the DCNN classification system trained on WSBWO is used to classify brain tumors. To validate the suggested work over alternative approaches, various studies were finally carried out.