Convolutional Neural Network-Based Brain Tumor Segmentation Using Detectron2

Brain tumors are a serious life-threatening disease that can generate tremendously negative impact in the patients lives. Thus, early detection is critical for successful treatment, however the complexity of the tumors which can vary in size, shapes and position complicate the task. For that reason this paper presents a convolutional neural network approach using the Mask R-CNN architecture through the Detectron2 framework in order to ease neurologists diagnosis and speed up the treatments. The model was trained on a dataset consisting of MRI images of brain tumors from two publicly available sources. The dataset was pre-processed and augmented to improve its accuracy. Our results show that the model achieved a high accuracy of 96.02% in segmenting brain tumors. This is a significant improvement over existing methods and can potentially lead to more accurate and timely diagnoses of brain tumors. Therefore, this work demonstrates the potential of convolutional neural networks models in medical imaging and the importance of accurate and timely detection of brain tumors. The model developed in this work can be used as a tool to aid clinicians in the diagnosis and treatment of brain tumors, ultimately improving patient outcomes.