Brain Tumor Detection System Using Hybrid Image Segmentation Techniques

Medical imaging plays a pivotal role in the detection and diagnosis of brain tumors, aiding clinicians in treatment planning and patient management. However, the inadequacies of using single image segmentation techniques for accurate tumor detection pose significant challenges. In this study, we propose a novel approach to enhance brain tumor detection by combining segmentation techniques from both Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) scans. By leveraging the complementary strengths of MRI in soft tissue contrast and CT in bone visualization, our method aims to improve the accuracy and reliability of tumor segmentation. Through extensive experimentation and evaluation on a dataset of brain tumor images, we achieved promising results. The Dice Coefficient, a measure of segmentation accuracy, yielded a value of 0.7063, indicating substantial agreement between predicted and ground truth segmentations. Furthermore, the Intersection over Union (IoU) metric reached 0.5460, demonstrating significant overlap between segmented and true tumor regions. Additionally, our model attained a Binary Accuracy of 0.7388, reflecting high correctness in classifying tumor and non-tumor regions. The validation tests also indicate effective optimization and generalization of the segmentation model. Overall, our study underscores the potential of integrating MRI and CT scan segmentation techniques to enhance brain tumor detection accuracy, thus facilitating more precise clinical decision making and improving patient outcomes.