ENHANCED FAST MASK R-CNN WITH SNAKE SWARM OPTIMIZATION FOR PRECISE BONE TUMOR SEGMENTATION AND CLASSIFICATION

Abstract

Bone tumor detection and segmentation in medical imaging are critical for accurate diagnosis and treatment planning. However, challenges such as tumor variability, image noise, and complex morphological features can hinder the performance of traditional segmentation methods. This paper presents an enhanced framework that integrates Fast Mask R-CNN, a state-of-the-art deep learning model for segmentation, with Snake Swarm Optimization (SSO) to improve the precision and efficiency of bone tumor segmentation and classification. The SSO algorithm, inspired by the hunting strategies of snakes, is employed to dynamically optimize key model parameters, such as anchor sizes, learning rates, and mask thresholds. This dynamic optimization allows the model to adapt to variations in tumor shapes and imaging conditions, enhancing segmentation accuracy and boundary delineation. Additionally, the SSO’s exploration-exploitation balance helps improve noise resilience, reducing artifacts and preserving important tumor features. Experimental validation on a diverse set of medical imaging datasets demonstrates that the integrated approach significantly outperforms baseline methods in segmentation accuracy, tumor boundary precision, and classification robustness. The proposed model holds strong potential for real-time clinical applications, providing radiologists with a powerful tool for early and accurate bone tumor detection, classification, and treatment planning