Identification and categorization of skin cancer using a Convolutional Neural Network
Authors: Archana Prajapati, Arpita Dash, Priya
Country: India
Full-text Research PDF File: View | Download
Abstract: Skin cancer is a prevalent form of cancer that poses significant health risks. It is crucial to detect this disease early, as with other cancers, to effectively manage treatment. Traditional methods of skin cancer diagnosis, however, tend to be inaccurate and can lead to unnecessary biopsies. Moreover, some existing machine learning models for cancer detection support only a limited number of skin cancer types, which can restrict their usefulness. This study developed a system using a Convolutional Neural Network capable of autonomously distinguishing between skin cancer and benign tumor lesions. The introduced model features three hidden layers with output channels scaling from 16, to 32, to 64. It employs several optimizers—SGD, RMSprop, Adam, and Nadam—with a learning rate of 0.001. Among these, the Adam optimizer yielded the highest accuracy at 93% for classifying skin lesions into benign or malignant categories using the ISIC dataset. These results outperform the current methods of skin cancer classification.
Keywords: Skin Cancer , ISIC , Convolutional Neural Network, Adam, and Nadam.
Paper Id: 230721
Published On: 2024-06-30
Published In: Volume 12, Issue 3, May-June 2024
Cite This: Identification and categorization of skin cancer using a Convolutional Neural Network - Archana Prajapati, Arpita Dash, Priya - IJIRMPS Volume 12, Issue 3, May-June 2024.