Cloud-Based Skin Cancer Diagnosis Using Convolutional Neural Networks with Efficientnet

Author(s): Priyadarshini Radhakrishnan, Yashwant Kumar Kolli, Vijai Anand Ramar, Karthik Kushala, Venkataramesh Induru and Thanjaivadivel M

Abstract:

The increasing incidence of skin cancer has called for the creation of early, accurate, and computerized diagnostic systems. Existing techniques often suffer from high error rates and limited preprocessing capabilities. To reverse this, work seeks to develop a deep learning model for the precise classification of skin disease from the International Skin Imaging Collaboration dataset for effective detection of malignant and benign lesions. The procedure starts by gathering skin cancer image data from the International Skin Imaging Collaboration dataset and performing preprocessing using Contrast Limited Adaptive Histogram Equalization and Gaussian filtering for noise removal and contrast enhancement of images are fed into a Convolutional Neural Network to receive features, subsequently, Efficient Net categorizes the inputs as non-cancer and cancer classes, and uploads non-cancer cases to a cloud storage system for convenient access and effective data management. Experimental tests validate the effectiveness of the system, where the Efficient Net classifier performs with better precision of 96.8%, accuracy of 95.4%, recall of 94.7%, and F1-score of 95%. The study offers a reliable framework that maximizes diagnostic reliability and availability since it is a notable contribution to intelligent healthcare solutions.

keywords:

Pages: 82-87     |    2 View     |    0 Download