Title: Classification of skin lesion images using proposed convolutional neural network

Authors: Hema Vastrakar; Akhilesh Kumar Shrivas; Amit Kumar Chandanan

Addresses: Department of Computer Science and Information Technology, Guru Ghasidas Vishwavidyalaya (Central University), Bilaspur, Chhattisgarh, India ' Department of Computer Science and Information Technology, Guru Ghasidas Vishwavidyalaya (Central University), Bilaspur, Chhattisgarh, India ' Department of Computer Science Engineering, Guru Ghasidas Vishwavidyalaya (Central University), Bilaspur, Chhattisgarh, India

Abstract: People around the world struggle with a wide variety of illnesses, including skin cancer, which is one of the most common. It originates in the skin, can spread to other organs, cause disfigurement, and can even be deadly. This research developed a tuned Convolutional Neural Networks (CNN) based model for the classification of image-based skin cancer diseases. The performance of the proposed CNN model is evaluated with different tuning parameters of CNN to achieve better classification accuracy. The models' behaviours vary because of the varying numbers of layers contained by CNN. The performance of the CNN model is compared with different numbers of epochs and activation functions. The performance of the proposed CNN model is also compared with that of the sigmoid, Tanh, and ReLU activation functions. This research work shows that our proposed sequential CNN model achieved good accuracy of 94.87% with 45 epochs in Relu activation function.

Keywords: classification; deep learning; machine learning; CNN; convolutional neural networks; skin cancer; activation function.

DOI: 10.1504/IJGUC.2024.140133

International Journal of Grid and Utility Computing, 2024 Vol.15 No.3/4, pp.380 - 395

Received: 01 Jun 2023
Accepted: 28 Apr 2024
Published online: 24 Jul 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article