What question did this study set out to answer?

This research aims to develop a robust and explainable AI framework for accurate skin lesion segmentation while addressing uncertainty in predictions.

June 5, 2026Machine Learning and Knowledge ExtractionOpen Access

UncerKAN-Mamba: A Clinically Robust, Transparent, and Explainable AI Framework for Low-Latency Skin Lesion Segmentation with Deterministic Single-Pass Uncertainty Estimation

View Full Paper

Ask AI

Bookmark

Authors

HKHüseyin Kutlu CÇCemil Çolak

Discussion

Member takes

Overview

Randomized trial evaluates low-latency skin lesion segmentation, indicating its clinical applicability.

Key Points

This research aims to develop a robust and explainable AI framework for accurate skin lesion segmentation while addressing uncertainty in predictions.
Integrated EfficientNet-B4 encoder with Mamba state space model blocks in a UNet++ decoder.
Utilized a Kolmogorov–Arnold Network for deterministic single-pass uncertainty estimation.
Applied Grad-CAM and other metrics for explainability assessment.
Achieved Dice scores of 0.8958, 0.9214, and 0.9360 on ISIC 2018, PH2, and ISIC 2016 datasets, respectively.
KAN uncertainty showed Pearson correlation of r = 0.674–0.731 with segmentation error, outperforming MC Dropout and Deep Ensembles.
Quantitative metrics indicated strong interpretability with AUROC = 0.971 and excellent calibration rates.

Cite This Study

Kutlu et al. (2026) studied this question.

synapsesocial.com/papers/6a22698b763171746d54824e https://doi.org/10.3390/make8060153

View Full Paper

Ask AI

Bookmark

June 5, 2026Machine Learning and Knowledge ExtractionOpen Access

UncerKAN-Mamba: A Clinically Robust, Transparent, and Explainable AI Framework for Low-Latency Skin Lesion Segmentation with Deterministic Single-Pass Uncertainty Estimation

View Full Paper

Ask AI

Bookmark

Authors

HKHüseyin Kutlu CÇCemil Çolak

Discussion

Member takes

Overview

Randomized trial evaluates low-latency skin lesion segmentation, indicating its clinical applicability.

Key Points

This research aims to develop a robust and explainable AI framework for accurate skin lesion segmentation while addressing uncertainty in predictions.
Integrated EfficientNet-B4 encoder with Mamba state space model blocks in a UNet++ decoder.
Utilized a Kolmogorov–Arnold Network for deterministic single-pass uncertainty estimation.
Applied Grad-CAM and other metrics for explainability assessment.
Achieved Dice scores of 0.8958, 0.9214, and 0.9360 on ISIC 2018, PH2, and ISIC 2016 datasets, respectively.
KAN uncertainty showed Pearson correlation of r = 0.674–0.731 with segmentation error, outperforming MC Dropout and Deep Ensembles.
Quantitative metrics indicated strong interpretability with AUROC = 0.971 and excellent calibration rates.

Cite This Study

Kutlu et al. (2026) studied this question.

synapsesocial.com/papers/6a22698b763171746d54824e https://doi.org/10.3390/make8060153

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark