EfficientSkinSegNet: a lightweight convolutional neural network for accurate skin lesion segmentation

Abstract

Accurate segmentation of skin lesions is crucial for the early detection and treatment of skin cancer. In this study, we propose EfficientSkinSegNet, a novel lightweight convolutional neural network architecture specifically designed for precise skin lesion segmentation. EfficientSkinSegNet incorporates efficient feature extraction encoders and decoders, leveraging multi-head convolutional attention and spatial channel attention mechanisms to extract and enhance informative features while eliminating redundant ones. Furthermore, a multi-scale feature fusion module is introduced in the skip connections to facilitate effective fusion of features at different scales. Experimental evaluations on benchmark datasets demonstrate that EfficientSkinSegNet outperforms state-of-the-art methods in terms of segmentation accuracy while maintaining a compact model size. The proposed network shows promise for practical clinical diagnostic applications, providing a balance between segmentation performance and computational efficiency. Future research will focus on evaluating EfficientSkinSegNet’s performance on diverse semantic segmentation tasks and optimizing it for medical image analysis.