Achievements of neural network in skin lesions classification

Hameed, Nazia and Shabut, Antesar and Hameed, Fozia and Cirstea, Silvia and Hossain, Alamgir (2021) Achievements of neural network in skin lesions classification. In: State of the Art in Neural Networks and their Applications. Academic Press, London, UK, pp. 133-151. ISBN 978-0-12-819740-0

Full text not available from this repository.
Official URL: https://doi.org/10.1016/B978-0-12-819740-0.00007-3

Abstract

The gross mismatch of skin disease cases and the specialties to manage them is the main cause of a continuously increased disease burden. The skin disease burden contributes 1.79% toward the global disease burden. To lessen this burden, automated skin lesions classification schemes that can provide multiclass classification are highly demanded. This chapter presents an investigation into an automated classification scheme to classify multiple skin lesions (acne, eczema, psoriasis; benign, and malignant) using state-of-the-art machine learning techniques. In the proposed classification scheme, convolution neural network (CNN) is utilized using the transfer learning approach, and a pretrained CNN model “AlexNet” is used to retrain the classification model on the skin lesion dataset. The proposed classification scheme outperformed over existing classification schemes and obtained an accuracy of 96.65%. The multiclass classification scheme can be very beneficial in the limited resource areas as it can assist in the early diagnosis of multiple skin lesions.

Item Type: Book Chapter
Keywords: Skin lesions classification, machine learning, automated skin diseases classification, skin cancer classification, acne classification, psoriasis classification, eczema classification, dermatological image classification, deep learning, transfer learning, computer-aided diagnosis
Faculty: Faculty of Science & Engineering
Depositing User: Lisa Blanshard
Date Deposited: 04 Apr 2022 09:32
Last Modified: 04 Apr 2022 10:13
URI: https://arro.anglia.ac.uk/id/eprint/707459

Actions (login required)

Edit Item Edit Item