DHIYANESH B,RAMKUMAR M,SHARMASTH VALI Y,ASHA A,SANGEETHAPRIYA R
DOI: https://doi.org/Effective patient evaluation and therapy depend on an understanding of brain tumors, especially their categorization. Although a variety of imaging methods are used to diagnose brain tumors, magnetic resonance imaging (MRI) is the most popular because it produces images with higher quality, less ionizing radiation, and more accuracy and precision. The usefulness of hybrid transfer learning models for identifying and categorizing brain cancers is investigated in this work. We introduce a novel hybrid transfer learning-based model with good sensitivity and accuracy for segmentation using Mask R-CNN (region-based convolutional neural network). Several metrics, such as F1 score, recall, specificity, sensitivity, precision, and overall accuracy, are used to thoroughly assess the model's performance. In our experiments, the proposed model was tested on three different architectures: VGG16, Inception V3, and ResNet-50. Remarkably, it achieved precision and sensitivity rates of over 99%, significantly outperforming the performance of current methodologies. The model was trained using Efficient-ResNet hybrid transfer learning architectures, which further contributed to its robustness and reliability. By integrating these advanced techniques, our approach not only improves tumor detection capabilities but also supports more accurate treatment planning for patients. The results show that hybrid transfer learning has the potential to revolutionize medical imaging, opening the door to better patient outcomes and more accurate diagnosis in the treatment of brain tumors. This work lays the groundwork for future research aimed at refining these models and exploring their applications in other areas of medical imaging.
