Leveraging CNNs and Transformers: The Hybrid Model for Brain Tumor Detection and Segmentation

Brain tumor, which are defined by abnormal cell growth, require urgent care and treatment, as if untreated, they can be fatal. The detection and segmentation are essential for a precise diagnosis and treatment planning. However, due to the intricate anatomy of the brain and the diversity of types of tumor, these tasks are extremely difficult. Radiologists use magnetic resonance imaging (MRI) because of its exceptional ability to identify even the smallest abnormalities in the brain. Recent advances in image processing and computer vision have led to significant advances in automated diagnosis in the field of healthcare technology, notably improving the accuracy, cost, and time efficiency of diagnosis. This paper proposes a hybrid method composed of Yolo11n and SAM for the tumor detection and segmentation. Yolo11n was used to detect the tumor, and then the Segmentation Anything model was utilised to segment the tumor. Using the advantages of these cutting-edge techniques, the hybrid model provides a reliable way to detect and segment MRI data accurately for automatic detection of brain tumors. The suggested method achieves a Dice score of 96.06% and an IoU Score of 92.43%, which is a substantial improvement over previous models in terms of segmentation performance. Furthermore, the detection model obtains mAP@50, precision, and Recall scores of 99.41%, 98.1%, and 98.41%, respectively.