DEEP LEARNING-BASED AUTOMATED SKIN CANCER DIAGNOSIS USING AN ENHANCED XCEPTION NETWORK
International Research Journal of Modernization in Engineering Technology & Science · 08.07.2026
DOI: 10.56726/irjmets101591 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Research Journal of Modernization in Engineering Technology & Science, online erschienen am 08.07.2026.
Abstract (Originalsprache)
Early identification of skin cancer is essential because timely diagnosis greatly increases the chances of successful treatment.Although dermoscopic examination is widely used by dermatologists, manual assessment may become difficult when different skin lesions exhibit similar visual characteristics.Recent progress in deep learning has encouraged the development of computer-assisted systems that support specialists by providing fast and reliable image classification.This study develops an automated skin lesion classification model by adapting the Xception convolutional neural network through transfer learning.Instead of training a network from the beginning, knowledge learned from ImageNet is fine-tuned for dermoscopic image analysis, reducing training time while preserving strong feature extraction capability.The HAM10000 dataset, which contains seven categories of skin lesions, is used to train and evaluate the model.Before training, the images undergo preprocessing operations such as resizing, normalization, duplicate removal, and augmentation to improve data quality and increase the diversity of the training samples.The trained model classifies dermoscopic images into Actinic Keratosis, Basal Cell Carcinoma, Benign Keratosis, Dermatofibroma, Melanoma, Melanocytic Nevi, and Vascular Lesions.Experimental evaluation produced an overall accuracy of 98.86%, with consistently high Precision, Recall, and F1-score, demonstrating reliable performance across multiple lesion categories.To improve usability, the trained model is connected to a Python Tkinter-based graphical interface that allows users to upload dermoscopic images and receive prediction results.The outcomes of this study indicate that transfer learning combined with effective preprocessing can provide an efficient decision-support tool for assisting dermatologists in skin lesion classification and promoting earlier clinical diagnosis.
Zitieren
APA
(2026). DEEP LEARNING-BASED AUTOMATED SKIN CANCER DIAGNOSIS USING AN ENHANCED XCEPTION NETWORK. International Research Journal of Modernization in Engineering Technology & Science. https://doi.org/10.56726/irjmets101591
Vancouver
DEEP LEARNING-BASED AUTOMATED SKIN CANCER DIAGNOSIS USING AN ENHANCED XCEPTION NETWORK. International Research Journal of Modernization in Engineering Technology & Science. 2026. doi:10.56726/irjmets101591
BibTeX
@article{euvimed2026,
title = {DEEP LEARNING-BASED AUTOMATED SKIN CANCER DIAGNOSIS USING AN ENHANCED XCEPTION NETWORK},
journal = {International Research Journal of Modernization in Engineering Technology & Science},
year = {2026},
doi = {10.56726/irjmets101591}
}
EUVIMED