Home / Current Issue / Paper 1715038
AI-Based Skin Disease Classification Using Deep Learning on Dermoscopic Images
Subject area: Science,Engineering and Technology · Area of research: Artificial Intelligence
Abstract
Skin diseases are one of the major health challenges in the world, and their occurrence differs among different populations. The health consequences of these diseases can be extremely severe. Therefore, early and precise detection of the diseases is very important in order to intervene in such cases, and this can be done in an accurate manner using the proposed model as compared to the conventional diagnosis process, as the conventional diagnosis process is more dependent on visual inspection and can differ according to the availability of dermatologists in different regions of the world. Therefore, this paper presents a novel framework based on AI technology to classify different skin diseases using dermoscopic images. Although the conventional diagnosis process can be more prone to error, the proposed model is based on a hybrid model that uses CNN and Transformer, and different images are used to train the model in order to make it more robust and accurate to handle images of different skin tones and types, as different images are used to train the model using various data sets, including ISIC and HAM10000, and the results show that the proposed ensemble learning approach can achieve more than 96Additionally, we incorporate Explainable AI (XAI) methods, including Grad- CAM and SHAP, to enable clinicians to interpret the rationale behind each classification decision through transparent visual explanations. By bridging the gap between high-performance deep learning and interpretability, this research targets the development of teledermatology initiatives and improves the accessibility of healthcare services in under-served regions.
References
[1] kshirsagar, P. R., et al. (2022). Deep Learning Approaches for Prognosis of Automated Skin Disease. [Custom real-time dataset study.
[2] Gao et al. (2023): Introduced a new AI text detection scheme for the exam setting, involving human-written answer classifiers based on the BERT system.
[3] Rudolph et al.: Their research focused on the influence of AI-generated text in education and the difficulties in identifying AI-generated texts in academic examinations.
[4] Clark et al. (2023) analyzed several recent instances of students using various AI tools to finish their programming assignments and proposed different approaches to detect AI-generated code.
[5] In “AI in Disease Diagnosis” from 2022, the researcher summarized various studies using artificial intelligence in medical diagnosis. They used various medical datasets like ADNI, UCI, ISIC, and MIMIC.
[6] Ravi proposed a deep learning for skin cancer detection considering misclassification costs. Experiments were carried out on the ISIC and HAM10000 datasets.
[7] Janbi et al. (2023) proposed Imtidad, a distributed artificial intelligence system for diagnostic assistance in skin diseases, providing validation on the dataset HAM10000.
[8] Tadesse et al. [24] proposed the STAR-ED framework that studies the representation of skin tone in educational datasets, using the datasets Fitzpatrick17K and SegmentedSkin to address issues of fairness.
How to cite this paper
@article{1715038,
author = {Balaji G, Monisha V K},
title = {AI-Based Skin Disease Classification Using Deep Learning on Dermoscopic Images},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {9},
pages = {3537-3545},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1715038.pdf},
abstract = {Skin diseases are one of the major health challenges in the world, and their occurrence differs among different populations. The health consequences of these diseases can be extremely severe. Therefore, early and precise detection of the diseases is very important in order to intervene in such cases, and this can be done in an accurate manner using the proposed model as compared to the conventional diagnosis process, as the conventional diagnosis process is more dependent on visual inspection and can differ according to the availability of dermatologists in different regions of the world. Therefore, this paper presents a novel framework based on AI technology to classify different skin diseases using dermoscopic images. Although the conventional diagnosis process can be more prone to error, the proposed model is based on a hybrid model that uses CNN and Transformer, and different images are used to train the model in order to make it more robust and accurate to handle images of different skin tones and types, as different images are used to train the model using various data sets, including ISIC and HAM10000, and the results show that the proposed ensemble learning approach can achieve more than 96Additionally, we incorporate Explainable AI (XAI) methods, including Grad- CAM and SHAP, to enable clinicians to interpret the rationale behind each classification decision through transparent visual explanations. By bridging the gap between high-performance deep learning and interpretability, this research targets the development of teledermatology initiatives and improves the accessibility of healthcare services in under-served regions.},
month = {March},
doi = {https://doi.org/10.64388/IREV9I9-1715038}
}