MixLVMM: A Mixture of Lightweight Vision Mamba Model for Enhancing Skin Lesion Segmentation Across High Tone Variability
Bibliographic record
Abstract
Accurate skin lesion segmentation remains a critical challenge in automated dermatological diagnosis due to heterogeneous lesion presentations, ambiguous boundaries, imaging artifacts, and significant variability in skin and lesion tones across diverse populations. Current segmentation methods inadequately address these multifaceted complexities, particularly failing to handle extreme tone variations that can lead to diagnostic bias. To address these limitations, we present the Mixture of Lightweight Vision Mamba Model (MixLVMM), a novel expert-based framework that enhances segmentation performance across high tone variability through specialized processing. Our approach employs a Siamese network with triplet loss as a gate mechanism to categorize lesions based on tonal characteristics, routing each image to specialized Vision Mamba Model (VMM) experts optimized for specific lesion categories. Each expert utilizes a U-shaped architecture incorporating Focused Vision Mamba blocks and Adaptive Salient Region Attention modules to capture lesion-specific features while maintaining computational efficiency. Comprehensive evaluation on ISIC and PH2 datasets demonstrates that MixLVMM achieves superior segmentation performance with an average Dice coefficient of 93%, surpassing state-of-the-art methods while maintaining efficiency with only 2.5M parameters. These results establish MixLVMM as a robust solution for addressing tone-related segmentation challenges in clinical dermatology, offering both high accuracy and practical deployment feasibility for real-world applications. Additional materials and code will be available at https://github.com/MOHAMEDLamine77/MixLVMM.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".