An Effective Detection of Skin Cancer Using A Multi-Module CNN

In modern years, Deep Learning (DL) systems, predominantly Convolutional Neural Networks (CNNs), have shown hopeful outcomes in ‎computer-aided diagnosis and classification of skin lesions. In this paper, an advanced methodology for skin cancer detection is proposed ‎that leverages preprocessing and classification-based multi-module CNN. The proposed method begins with preprocessing and data ‎augmentation steps for improving the resolution of the skin lesions, including noise reduction and contrast enhancement, which optimize the ‎input data for subsequent analysis. Subsequently, a multi-module CNN architecture is employed, consisting of interconnected modules ‎designed to capture diverse features crucial for accurate classification. The overall proposed work is implemented in Python software, and ‎a comparative analysis is carried out with the existing approaches to show the prominence of the developed work. The results of the ‎investigation show that the model created has a high accuracy of 97% and a short classification time, which makes it a useful tool for ‎helping dermatologists detect skin cancer early‎.