Machine Learning and Deep Learning Methods for Skin Lesion Classification and Diagnosis: A Systematic Review

Area of Science:

• Dermatology
• Medical Imaging
• Artificial Intelligence

Background:

• Computer-aided diagnosis (CAD) systems are increasingly researched for skin lesion analysis.
• Evaluating the diagnostic accuracy of these systems is crucial.

Purpose of the Study:

• To review, synthesize, and evaluate evidence on the diagnostic accuracy of computer-aided skin lesion diagnosis systems.
• To compare traditional machine learning and deep learning methods in this field.

Main Methods:

• Systematic review of articles published in the last five years from ScienceDirect, IEEE, and SpringerLink.
• Analysis of 53 articles using traditional machine learning and 49 using deep learning methods.

Main Results:

• Comparison of studies based on contributions, methodologies, and outcomes.
• Identification of common challenges in evaluating skin lesion segmentation and classification.

Conclusions:

• The review highlights the need for robust evaluation metrics and larger, diverse datasets.
• Addressing issues like small datasets, ad hoc image selection, and racial bias is essential for improving CAD systems for skin lesions.