Authors: E. Priyadharshini
Abstract: Medical image analysis plays a significant role in the early detection and diagnosis of cancer. Accurate segmentation and classification of malignant tissues are essential for improving clinical decision-making and patient outcomes. In recent years, Artificial Intelligence (AI) and Machine Learning (ML) techniques, particularly neural networks, have demonstrated remarkable success in biomedical image processing applications. However, the performance of conventional Multi-Layer Perceptron (MLP) networks is highly dependent on optimal parameter tuning, which remains a challenging task due to the complexity and high dimensionality of medical image data. This paper proposes an optimized MLP model using the Salp Swarm Algorithm (SSA) for malignant tissue segmentation and classification in biomedical images. SSA is a nature-inspired metaheuristic optimization technique modeled on the swarming behavior of salps in ocean environments. The algorithm offers strong global search capability, faster convergence, and improved avoidance of local optima compared with traditional optimization methods. By integrating SSA with the MLP network, the proposed model enhances feature selection, weight optimization, and classification accuracy. The proposed SSA-MLP framework is evaluated using publicly available biomedical image datasets. Performance assessment is carried out using standard evaluation metrics including Accuracy, Sensitivity, Specificity, Precision, F1-Score, and Area Under the Receiver Operating Characteristic Curve (AUC-ROC). Experimental results demonstrate that the SSA-tuned MLP model achieves superior performance when compared with conventional machine learning and neural network approaches. The model shows improved segmentation quality, enhanced classification capability, and greater robustness in detecting malignant tissues. This study contributes to the advancement of intelligent medical imaging systems by presenting a reliable and efficient optimization-based neural network model for cancer diagnosis. The findings indicate that SSA can significantly improve neural network performance in medical image analysis, thereby supporting accurate diagnosis and effective clinical decision support systems.curve Receiver Operating Characteristic (AUC-ROC)
