Detection and classification of melanoma using convolutional neural networks (CNNs) on dermoscopy images
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
The fast growth and high death rate of melanoma, an extremely aggressive skin cancer, make it a major danger to world health if not caught early. Dermoscopy pictures provide for a more precise diagnosis by providing a non-invasive way to see skin lesions. This study introduces a method for detecting and classifying melanoma in dermoscopy pictures that is based on deep learning and uses Convolutional Neural Networks (CNNs). The suggested model accurately distinguishes between malignant and benign skin lesions by using a convolutional neural network (CNN) architecture to extract important characteristics from dermoscopic data. The research makes use of a freely accessible dataset of dermoscopy pictures that have been preprocessed using data augmentation methods to lessen the likelihood of overfitting and improve the generalisability of the models. Important measures including accuracy, sensitivity, specificity, and F1-score were used to optimise the CNN architecture via hyperparameter tweaking. The model's efficacy is shown by experimental findings, which show that it achieves an accuracy of over 95%, has great sensitivity, and is very specific, guaranteeing early detection with minimised false positives. The results show that convolutional neural network (CNN) techniques have promise as an automated, dependable tool for melanoma diagnosis, which might improve dermatological diagnostics. By paving the way for scalable, precise, and economical methods for early cancer diagnosis, this study highlights the significance of AI in transforming healthcare.
Fields of Study & Subjects
Cutaneous Melanoma Detection and Management
AI in cancer detection
Advanced Neural Network Applications
Publication details
- Journal / source
- AIP conference proceedings
- Volume
- 3401
- Pages
- 020137–020137
- Publication date
- 1 January 2026
- Language
- EN