ODNN: Augmented Brain Tumor Boundary Detection and Localization in MRI Images using Optimization Guided Hybrid Deep Neural Networks

Accurate tumor segmentation and localization in MRI images are critical for diagnosis, treatment planning, and monitoring in oncology. However, manual delineation of tumor boundaries is time-consuming and subject to inter-observer variability. Deep learning techniques have shown promise in automating this process, yet achieving precise localization remains challenging. In this work, we propose an Optimization-Guided Hybrid Deep Neural Network (ODNN) framework for augmented tumor boundary detection and localization in MRI images. The ODNN model integrates optimization techniques practical swam optimization (PSO) within Inception-v3 and ResNet-50 to enhance the accuracy of tumor localization. Specifically, we employ a combination of convolutional neural networks (CNNs) for feature extraction and optimization algorithms to refine tumor boundary predictions. The optimization technique, PSO process iteratively adjusts model parameters to minimize a predefined tumor function, optimizing the network for improved localization performance. We evaluate the proposed ODNN framework on a dataset of MRI images containing various tumor types and complexities. Comparative experiments demonstrate that our approach achieves superior performance in tumor boundary detection and localization compared to baseline deep learning models. Quantitative evaluation metrics such as precision of 97.8%, dice similarity coefficient of 94.5%, recall of 95.5%, f1-score of 96.2% and hausdorff distance of 7.12 which confirm the effectiveness of the ODNN framework in accurately delineating tumor boundaries. By integrating optimization techniques with deep learning, we effectively address the challenges of precise tumor localization. The proposed ODNN framework holds promise for improving diagnostic accuracy, treatment planning, and patient outcomes in oncology, paving the way for advancements in computer-aided diagnosis systems for medical imaging.