CAMEO: A Class-Activation-Mapped Equitable Overlay Framework for Fair and Robust Deep Learning-based Skin Condition Diagnosis

Youssef Attia; Debasmita Mukherjee

eess.IVSep 28, 2026

CAMEO: A Class-Activation-Mapped Equitable Overlay Framework for Fair and Robust Deep Learning-based Skin Condition Diagnosis

Authors: Youssef Attia, Debasmita Mukherjee

Organizations: Department of Electrical and Computer Engineering, University of New Brunswick, Fredericton, NB, Canada

Abstract

Deep learning classifiers for dermoscopic skin lesions often reach high in-distribution accuracy while quietly relying on spurious background cues such as skin tone, device vignetting, and embedded rulers, rather than on lesion morphology. This undermines robustness and fairness across skin tones. This work asks whether Explainable AI (XAI), typically used only to audit a finished model, can instead be repurposed as an active training signal that corrects this shortcut without sacrificing diagnostic accuracy. We introduce CAMEO (Class Activation Mapped Equitable Overlay), a framework that improves skin-lesion classification by selecting stable model explanations and using them to separate lesions from their backgrounds. It then replaces the background with realistic synthetic skin while keeping the lesion unchanged. On HAM10000 and dark-skin ISIC images, CAMEO maintained accuracy while reducing background-driven errors by nearly four times. It also made the model's attention more consistent when backgrounds changed. Results across multiple tests show that reducing reliance on background information improves robustness, with Fitzpatrick-based backgrounds providing a realistic and interpretable approach. Results show that XAI-guided augmentation can make dermoscopic classifiers measurably more robust and fair at no cost to accuracy. They also clarify that it is the mechanism and not the specific tone palette that matters, and that the lasting contribution of XAI here lies in stability-screened, annotation-free lesion localisation rather than in the robustness number itself.

Figures & tables

Set	$N$	Description and purpose
T0	342	Frozen class-balanced HAM10000 test set (in-distribution).
T1	100	External dark-skin ISIC set (real cross-tone shift).
T_swap	342	Synthetic background swap of T0 (controlled robustness).

Table 1: Dataset notation used throughout. T1 (external, real) and T_swap (synthetic, controlled) are different sets and are never compared directly. Two probes use T_swap-style compositing: the decision-flip and attention probe (Section 5.6 ) re-composites 150 lesions onto six tones, whereas the T_swap accuracy metric uses one fixed swap per T0 image (

N{=}342

Name	Core	Background strategy	Role
Base-Skewed (M1)	Full HAM10000, imbalanced (6,224 and 774)	None (raw images)	Naive baseline; reflects the natural class imbalance.
Base-Balanced (M2)	Balanced subset (774 and 774)	None (raw images)	Primary baseline to beat; class imbalance removed by undersampling.
CAMEO (ours) (M3_AugB_M2, v4)	Balanced subset (774 and 774)	ChromaSwap , symmetric on both classes	Headline model; the primary result reported throughout the paper.
CAMEO-Skewed (M3_AugB_M1)	Full HAM10000, imbalanced core	ChromaSwap , minority class only	Tests whether augmentation alone can substitute for explicit class balancing.
CAMEO-RandomBG	Balanced subset (774 and 774)	Random, non-Fitzpatrick background	Internal control: isolates whether Fitzpatrick-specific tones are necessary.
BiasAdv [ 26 ]	Balanced subset (774 and 774)	Bias-conflicting background against a frozen biased classifier	External debiasing baseline, named as in its source publication.

Table 2: Model and baseline glossary. Every name used in the paper is defined here in one place. “Core” is the training data recipe and “Background strategy” is what replaces the region outside the lesion mask, if anything. Identifiers from earlier pipeline versions are given in parentheses for traceability.

Panel of Figure 1 — Figure 1: Overview of the CAMEO pipeline. Top row: data preparation, XAI extraction and the SSIM stability screen that decides which explanations may be acted on and how strongly each sample is augmented. Middle row: ruler artifact suppression and composite mask generation. Bottom row: ChromaSwap augmentation, retraining and evaluation. The screen is a gate, not a diagnostic: no attribution map reaches the masking stage without passing through it.

Model	Accuracy	Macro F1	AUC-ROC	Cohen’s $\kappa$
Base-Skewed	0.743 [0.696, 0.789]	0.736	0.819 [0.774, 0.862]	0.485
Base-Balanced	0.848 [0.810, 0.886]	0.848	0.912 [0.881, 0.941]	0.696
CAMEO (ours)	0.836 [0.795, 0.874]	0.836	0.913 [0.880, 0.942]	0.673
CAMEO-Skewed	0.795 [0.751, 0.836]	0.791	0.898 [0.862, 0.931]	0.591

Table 3: Clean accuracy and related metrics on the frozen balanced test set T0 (

N=342

). Because T0 is class-balanced, accuracy equals balanced accuracy, so a single accuracy column is reported. Bracketed values are bootstrap 95% confidence intervals (10,000 resamples) for the adjacent metric.

Model	Train	Val.	Test (T0)	Train $-$ Test
Base-Skewed	0.643	0.735	0.743	$-$ 0.100
Base-Balanced	0.981	0.795	0.848	$+$ 0.133
CAMEO (ours)	0.995	0.824	0.836	$+$ 0.159
CAMEO-Skewed	0.992	0.759	0.795	$+$ 0.196

Table 4: Train, validation and test accuracy (overfitting check). Train accuracy is measured on a random 2,000-image subset with evaluation transforms only. All values are accuracies.

Figure 2: Training and validation loss and accuracy against epoch for the augmented models, showing convergence and the observed gap between training and validation performance.

Statistic	Value
Images analysed	3,000
Mean SSIM	0.758
Median SSIM	0.766
SSIM range (min to max)	0.451 to 0.967
XAI-unstable (SSIM $<0.7$ )	744 (24.8%)
Background-bias flagged	1,559 (52.0%)

Table 5: XAI-instability statistics over the 3,000-image extraction set. SSIM is the mean structural similarity of the GradCAM map under three Gaussian input perturbations (

\sigma=0.01

Set	$\boldsymbol{N}$	Mean $R^{2}$	$\boldsymbol{R_{\mathrm{shift}}}$	ID rate
HAM10000 val. (calibration)	1,498	0.966	0.000	95.0%
Frozen T0 (control)	342	0.965	0.131	93.3%
T1 ISIC dark (external)	100	0.929	1.751	63.0%

Table 6: Covariate-shift statistics in the Base-Balanced feature space (tap 4).

R^{2}

is the per-image reconstruction R-square in the HAM10000 training PCA basis,

R_{\mathrm{shift}}

is the mean normalised per-feature shift, and the in-distribution (ID) rate is the fraction of images with

R^{2}\geq\tau=0.9426

Figure 3: $R^{2}$ distributions for the HAM10000 validation split, the frozen T0 test set and the external T1 dark-skin set. The vertical dashed line marks the in-distribution threshold $\tau=0.9426$ . T0 aligns with HAM10000 validation, whereas T1 shows a pronounced low- $R^{2}$ tail that corresponds to out-of-distribution images.

Model	Accuracy	Accuracy	$\boldsymbol{\Delta}$	AUC
	(all)	(ID)		(ID)
Base-Skewed	0.820	0.857	$+$ 0.037	0.861
Base-Balanced	0.640	0.810	$+$ 0.170	0.796
CAMEO (ours)	0.660	0.794	$+$ 0.134	0.676
CAMEO-Skewed	0.630	0.778	$+$ 0.148	0.726

Table 7: Accuracy on T1 (external dark-skin ISIC set): full set versus in-distribution (ID) subset (

n_{\text{ID}}=63

n_{\text{OOD}}=37

\Delta

is the accuracy gain obtained by restricting to ID images.

Model	Prob. range	Flip rate	CAM SSIM
Base-Balanced	0.250	19.8%	0.720
CAMEO (ours)	0.123	5.4%	0.842

Table 8: Background-counterfactual robustness. Decision stability: lower probability range and lower flip rate are better (150 images, 6 tones). Attention stability: higher GradCAM++ map SSIM before versus after a background swap is better (120 images).

Figure 4: Background shortcut proof on three lesions. Columns 1 to 3: original image with the ground-truth (GT) class and each model’s melanoma probability. Column 4: the same lesion with only the background swapped to a dark synthetic tone. Columns 5 and 6: predictions after the swap, green for stable or correct and red for a flip or error. Base-Balanced flips when the background darkens; CAMEO does not.

Figure 5: GradCAM++ attention on the same lesion with its original and a dark synthetic background. Base-Balanced (columns 2 and 3) drifts onto the background when the tone changes, giving a low map SSIM, whereas CAMEO (columns 4 and 5) stays on the lesion. Per-image SSIM is printed beneath each pair.

Figure 6: Decision stability under background swap. Left: per-image melanoma-probability range across tones, lower being more invariant. Right: mean probability standard deviation and label-flip rate. Bottom: a lesion whose probability under Base-Balanced swings widely with background tone while CAMEO stays nearly constant.

Group	$\boldsymbol{N}$	Mean IoU	Mean Dice	Median Dice
All	2,332	0.422	0.570	0.611
Non-melanoma	1,858	0.428	0.576	0.618
Melanoma	474	0.397	0.545	0.577

Table 9: Composite XAI mask versus the HAM10000 ground-truth lesion mask (ISBI 2018; Tschandl et al. [ 39 ] ). IoU is Intersection over Union.

Method	Mean IoU
Otsu (grayscale) [ 36 ]	0.552
Composite XAI mask (ours)	0.427
HSV colour threshold	0.426
SkinSAM without prompt [ 27 ]	$\approx$ 0.16

Table 10: Unsupervised segmentation baselines against ground truth (

N{=}2{,}332

). Otsu achieves higher raw Intersection over Union (IoU) on this light-skinned benchmark, but it relies on an intensity assumption that weakens on darker skin and it does not isolate the model-attributed lesion interior that ChromaSwap augmentation needs.

Figure 7: Qualitative validation of composite XAI masks against HAM10000 ground-truth lesion segmentations. The green outline is the clinical lesion mask and the heatmap overlay is the binary XAI mask used for ChromaSwap augmentation.

Figure 8: Exhaustive ChromaSwap augmentation across Fitzpatrick I to VI (rows), with three examples each (columns). The same lesions are carried through every row, so the effect of tone level reads down each column. The lesion interior is preserved while the background is substituted with a textured tone. Hex codes and labels are shown at the left.

Checkpoint	Pipeline / recipe	T0 accuracy	T_swap accuracy
Legacy-LesionJitter	comparison / unbalanced $+$ LesionJitter	0.798	0.792
Legacy-ChromaSwap	comparison / unbalanced $+$ ChromaSwap	0.827	0.810
Legacy-Combined	comparison / unbalanced $+$ A $+$ B	0.827	0.792
Legacy-ChromaSwap-p70	threshold sweep	0.825	0.807
Legacy-ChromaSwap-p80	threshold sweep	0.807	0.789
Legacy-FlatBG	v2 flat background	0.787	0.564

Table 11: Legacy augmentation ablations (v2, v3 and threshold-sweep checkpoints; these are not the headline CAMEO v4 model, see the text). All checkpoints are evaluated on the frozen T0 and T_swap manifests without retraining. T_swap is the synthetic background-counterfactual set and not the external T1 set.

Model	T0 accuracy	T_swap accuracy	Accuracy drop
Base-Balanced	$0.827\pm 0.016$	$0.745\pm 0.031$	$0.082\pm 0.035$
CAMEO (ours)	$0.836\pm 0.008$	$\mathbf{0.815\pm 0.017}$	$\mathbf{0.021\pm 0.022}$

Table 12: Multi-seed variance. Mean

\pm

standard deviation of accuracy over five seeds on the frozen T0 and T_swap manifests, where T_swap is the synthetic background-counterfactual set of Table 1 and not the external T1 set. “Accuracy drop” is the T0 to T_swap accuracy loss.

Model	Method	T0 accuracy	T_swap accuracy	$p$ vs. ours
CAMEO (ours)	XAI ChromaSwap (Fitzpatrick)	0.836	0.810	n/a
CAMEO-RandomBG	Random non-Fitzpatrick ChromaSwap	0.830	0.816	0.87
BiasAdv	BiasAdv-style conflicting background [ 26 ]	0.830	0.798	0.62
AdvDebias-GRL	Adversarial tone debiasing by gradient reversal [ 25 ]	0.833	0.734	$<$ 0.01

Table 13: Head-to-head debiasing baselines, all evaluated on the same T_swap manifest. “CAMEO (ours)” is the v4 headline checkpoint of Table 3 ; the three baselines are single-seed (42) retrains on the same balanced core with identical hyper-parameters. The last column is the two-sided McNemar

p

value against ours.

Figure 9: Qualitative pipeline results on three representative examples (rows). From left to right: original image, ruler-corrected GradCAM++ heatmap, cleaned composite mask (GradCAM++ plus Integrated Gradients, $p_{60}$ ), and the ChromaSwap augmented image with a Fitzpatrick-scale background.

Model	Clean acc.	$\boldsymbol{\varepsilon{=}.001}$	$\boldsymbol{\varepsilon{=}.01}$	$\boldsymbol{\varepsilon{=}.03}$	$\boldsymbol{\varepsilon{=}.05}$	$\boldsymbol{\varepsilon{=}.10}$
Base-Skewed	0.743	0.345	0.158	0.213	0.447	0.500
Base-Balanced	0.848	0.213	0.085	0.105	0.377	0.500
CAMEO (ours)	0.836	0.325	0.193	0.219	0.415	0.500
CAMEO-Skewed	0.795	0.409	0.272	0.289	0.386	0.503

Table 14: FGSM adversarial accuracy on the frozen T0 test set (

N=342

). Epsilon values are given in normalised input space. All values are accuracies.

Explore similar work

Uncertainty-Aware and Explainable Ensemble Deep Learning Framework for Multi-Class Skin Lesion Classification
Aug 11, 2026Rofiqul Islam, Lilatul FerdouseSkin Lesion Classification Melanoma
Department of Computer Science and Physics, Wilfrid Laurier University, Waterloo, ON, Canada
LLM-Based Visual Explanation Evaluation Framework for Assessing the Explainability of Facial Skin Disease Classification Models
Jun 15, 2026Gyuyeon NaSkin Lesion Classification Medical Vision-Language Models
AI and Business Analytics, Ewha Womans University, Seoul, Republic of Korea
PEFT-MedSAM: Efficient Fine-Tuning of Medical Foundation Models for Explainable Skin Lesion Segmentation
Jun 17, 2026Asad Channa, Abdullah Khan, Asghar Ali Chandio +4Skin Lesion Classification Lesion Segmentation
Department of Computer Science, Quaid-e-Awam University of Engineering, Sciences & Technology · Department of Artificial Intelligence, Quaid-e-Awam University of Engineering, Sciences & Technology, Pakistan · Department of Computer Science, Sindh Madressatul Islam University, City Campus, Karachi +1

Aug 11, 2026, eess.IVJ/K move · Enter open · S save

Uncertainty-Aware and Explainable Ensemble Deep Learning Framework for Multi-Class Skin Lesion Classification

Rofiqul Islam, Lilatul Ferdouse

Department of Computer Science and Physics, Wilfrid Laurier University, Waterloo, ON, Canada

Skin Lesion Classification Melanoma Roc-Auc Ensemble Intra-Class

Open paper AskPDF arXiv

Skin cancer diagnosis from dermoscopic images remains challenging due to high intra-class variability, inter-class similarity, class imbalance, and the limited interpretability of deep learning models. This paper proposes an uncertainty-aware and explainable deep learning framework for multi-class skin lesion classification. The framework combines a vision transformer model (MaxViT-Tiny) with CNN-based models (ConvNeXt-Tiny and EfficientNetV2-B0) through deep ensemble learning. Monte Carlo (MC) Dropout estimates predictive uncertainty and identifies unreliable predictions, while Grad-CAM++, an explainable AI (XAI) technique, provides visual explanations by highlighting lesion regions that influence model decisions. Evaluated on the HAM10000 dataset, the framework achieves 96% accuracy and 99% ROC-AUC under uncertainty-aware filtering (entropy < 1.0, confidence >= 0.7), with macro-average precision, recall, and F1-score of 94%, 95%, and 95%, respectively, and 96% weighted-average scores across all three metrics. The results demonstrate accurate, interpretable, and uncertainty-aware skin lesion classification for trustworthy computer-aided diagnosis.

Abstract

Explore similar work

Uncertainty-Aware and Explainable Ensemble Deep Learning Framework for Multi-Class Skin Lesion Classification

LLM-Based Visual Explanation Evaluation Framework for Assessing the Explainability of Facial Skin Disease Classification Models

PEFT-MedSAM: Efficient Fine-Tuning of Medical Foundation Models for Explainable Skin Lesion Segmentation