Brain Tumor Detection using RCNN and MobileNet

This study examines the identification and categorization of brain tumors by analyzing medical images. Based on their origin, it is said that brain tumors, which are an uncontrolled cell growth within the brain, can be classified into two broad categories. One is primary and the other is secondary. The main tumor detection technique is Magnetic Resonance Imaging (MRI), although it is only effective for tumors larger than 3 mm. Segmentation is frequently utilized for precise analysis of micro-sized cancers less than 3mm. To extract the features and perform classification, a method is suggested in this research which applies deep learning techniques such as Recurrent Convolutional Neural Networks and the Gray Level Co-occurrence Matrix. The improved K-means clustering technique is opted for image segmentation. The trials used a Kaggle dataset with 2870 training sets and 394 testing sets of MR images. The outcomes demonstrate that the suggested method outperforms earlier techniques like BP and U-Net. The study investigates the use of the MobileNet algorithm for classifying brain tumors in addition to the suggested technique, which yielded equivalent results. Overall, reliable and effective brain tumor detection and classification appear to be possible because of the combination of deep learning approaches with effective designs.