Explainable AI Framework for Precise and Trustworthy Skin Cancer Diagnosis

Vandana Kate, Arohi Kate, Chanchal Bansal, Charu Pancholi, Ashvini Patidar

2025

Abstract

Skin cancer, most especially melanoma, is a recognized health issue across the world and its management depends on early and correct diagnosis.Conventional methods like biopsies are relatively precise and reliable but they are time consuming and invasive and may cause either an infection or an outbreak. Non-invasive procedures such as dermoscopy depend on the knowledge of the physician, which can cause variability and randomness. To address these challenges, we propose an explainable AI (XAI) framework for precise and trustworthy skin cancer diagnosis. Our model integrates VGG16, InceptionV3, Inception-ResNet V2 and DenseNet-201 deep learning architectures fine-tuned on the HAM10000 benchmark dataset to distinguish skin lesions as benign or malignant. To ensure transparency and trust in the model’s predictions, we incorporate cutting-edge explainability techniques, including LIME (Local Interpretable Model-agnostic Explanations), SHAP (SHapley Additive exPlanations) and gradient-based methods like Grad-CAM. These tools highlight key image features and regions that influence model decisions. This proposed work deepens the knowledge in the field of using AI in the diagnosis of skin cancer and paves the way for integrating explainability into AI healthcare systems, improving accuracy and user trust.

Paper Citation

in Harvard Style

Kate V., Kate A., Bansal C., Pancholi C. and Patidar A. (2025). Explainable AI Framework for Precise and Trustworthy Skin Cancer Diagnosis. In Proceedings of the 3rd International Conference on Futuristic Technology - Volume 2: INCOFT; ISBN 978-989-758-763-4, SciTePress, pages 260-267. DOI: 10.5220/0013590500004664

in Bibtex Style

@conference{incoft25,
author={Vandana Kate and Arohi Kate and Chanchal Bansal and Charu Pancholi and Ashvini Patidar},
title={Explainable AI Framework for Precise and Trustworthy Skin Cancer Diagnosis},
booktitle={Proceedings of the 3rd International Conference on Futuristic Technology - Volume 2: INCOFT},
year={2025},
pages={260-267},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013590500004664},
isbn={978-989-758-763-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 3rd International Conference on Futuristic Technology - Volume 2: INCOFT
TI - Explainable AI Framework for Precise and Trustworthy Skin Cancer Diagnosis
SN - 978-989-758-763-4
AU - Kate V.
AU - Kate A.
AU - Bansal C.
AU - Pancholi C.
AU - Patidar A.
PY - 2025
SP - 260
EP - 267
DO - 10.5220/0013590500004664
PB - SciTePress