Volume 1 number 2 (03)

SKIN CANCER DETECTION: LEVERAGING CONVOLUTIONAL NEURAL NETWORKS FOR BINARY CLASSIFICATION

Pages 61-68

DOI 10.61552/JAI.2024.02.003

ORCID Amritanjali Swaroop, ORCID A. Charan Kumari, ORCID K. Srinivas


Abstract Globally Skin cancer accounts for millions of new cases annually posing a significant threat to public health, emphasizing the importance of timely detection to enhance patient outcomes and decrease mortality rates. In this research article an AI convolutional neural network (CNN) model for the detection and classification of skin lesions into benign and malignant type is presented. A comprehensive dataset comprising 33,126 dermoscopic images from the ISIC 2020 challenge, a tailored CNN architecture was developed and validated for accurate lesion classification. The good performance of the developed CNN model on various benchmark metrics highlight its capacity to precisely distinguish between benign and malignant skin lesions.The outcomes of this investigation indicate that the developed CNN model has the potential to significantly enhance diagnostic protocols by minimizing manual errors and streamlining patient care pathways through automation. Through the amalgamation of sophisticated machine learning methodologies with clinical protocols, the model not only supports immediate clinical implementations but also lays the groundwork for future advancements by incorporating larger datasets and real-world clinical information. This study provides a scalable CNN model for building an automated tool that can integrate into healthcare systems to facilitate early detection and treatment of skin cancer,resulting in cost reductions and improved health outcomes. This endeavor makes a remarkable impact in the realm of dermatology by emphasizing the utilization of artificial intelligence to transform the diagnosis of skin cancer and enhance patient care.

Keywords: Skin Cancer classification, Convolutional Neural Networks, Dermoscopic images, Early detection, Melanoma Classification.

Recieved: 11.02.2024. Revised: 21.03.2024. Accepted: 17.05.2024.

Publication Information

Publisher

Editor-in-Chief Director-in-Charge Managing Editor
Aleksandar Djordjevic

Frequency
Quarterly
Online ISSN

Indexing and Abstracting