Adaptive Fusion Boundary-Enhanced Multilayer Perceptual Network (FBAIM-Net) for Enhanced Polyp Segmentation in Medical Imaging

Accurate polyp segmentation in medical image analysis is vital for early diagnosis and treatment planning, particularly in scenarios with diverse polyp shapes and sizes. This study introduces the Adaptive Fusion Boundary-Enhanced Multilayer Perceptual Network (FBAIM-Net), leveraging a multi-attention mechanism and edge prediction for improved feature learning and segmentation performance. Experimental evaluations on challenging polyp datasets demonstrate FBAIM-Net’s superior performance over state-of-the-art methods, supported by quantitative metrics and qualitative analyses. FBAIM-Net presents a promising approach to advancing polyp segmentation in medical image analysis.