MambaLiteUNet: Cross-Gated Adaptive Feature Fusion for Robust Skin Lesion Segmentation

arXiv cs.CV / 4/23/2026

📰 NewsTools & Practical UsageModels & Research

共有:

Key Points

The paper introduces MambaLiteUNet, a compact skin lesion segmentation framework that embeds Mamba state space modeling within a U-Net backbone to better capture boundary and texture details.
It proposes three modules—Adaptive Multi-Branch Mamba Feature Fusion (AMF), Local-Global Feature Mixing (LGFM), and Cross-Gated Attention (CGA)—to strengthen local–global feature interaction and improve skip-connection quality.
On ISIC2017, ISIC2018, HAM10000, and PH2, MambaLiteUNet reportedly reaches an average IoU of 87.12% and Dice of 93.09%, outperforming prior state-of-the-art methods.
Compared with a standard U-Net baseline, the model improves average IoU by 7.72 points and Dice by 4.61 points while dramatically reducing parameters by 93.6% and GFLOPs by 97.6%.
In domain generalization to six unseen lesion categories, it achieves 77.61% IoU and 87.23% Dice, with results claimed to be best among evaluated models; the code is publicly available on GitHub.

Abstract

Recent segmentation models have demonstrated promising efficiency by aggressively reducing parameter counts and computational complexity. However, these models often struggle to accurately delineate fine lesion boundaries and texture patterns essential for early skin cancer diagnosis and treatment planning. In this paper, we propose MambaLiteUNet, a compact yet robust segmentation framework that integrates Mamba state space modeling into a U-Net architecture, along with three key modules: Adaptive Multi-Branch Mamba Feature Fusion (AMF), Local-Global Feature Mixing (LGFM), and Cross-Gated Attention (CGA). These modules are designed to enhance local-global feature interaction, preserve spatial details, and improve the quality of skip connections. MambaLiteUNet achieves an average IoU of 87.12% and average Dice score of 93.09% across ISIC2017, ISIC2018, HAM10000, and PH2 benchmarks, outperforming state-of-the-art models. Compared to U-Net, our model improves average IoU and Dice by 7.72 and 4.61 points, respectively, while reducing parameters by 93.6% and GFLOPs by 97.6%. Additionally, in domain generalization with six unseen lesion categories, MambaLiteUNet achieves 77.61% IoU and 87.23% Dice, performing best among all evaluated models. Our extensive experiments demonstrate that MambaLiteUNet achieves a strong balance between accuracy and efficiency, making it a competitive and practical solution for dermatological image segmentation. Our code is publicly available at: https://github.com/maklachur/MambaLiteUNet.