Optimizing Liver Tumor Segmentation in CT Scans: A Genetic Algorithm Approach for Low-Resource Medical Imaging

Abstract

Liver tumor segmentation is a crucial step in medical image analysis, aiding in the diagnosis and treatment of hepatic diseases. Traditional segmentation methods struggle with irregular tumor shapes, varying intensities, and low-contrast boundaries. Deep learning approaches, while effective, require large annotated datasets and high computational resources. The proposed work introduces an evolutionary Genetic Algorithm (GA)-based approach for liver tumor segmentation in CT scans. The GA optimizes a population of segmentation masks using an energy-based fitness function, evolving towards the best tumor segmentation boundaries. Experimental results demonstrate that our GA-based method achieves comparable accuracy to deep learning models while reducing computational time by 50%. The approach is particularly beneficial for low-resource environments and datasets with limited annotations.