Artificial intelligence for melanoma diagnosis: from support vector machines to deep learning and beyond

No Thumbnail Available
File version
Author(s)
Byth, LA
Byth, JL
Griffith University Author(s)
Primary Supervisor
Other Supervisors
Editor(s)
Date
2021
Size
File type(s)
Location

Virtual

License
Abstract

Melanoma diagnosis is a vexing problem, with practitioners requiring years of training and experience to achieve proficiency. Machine learning represents a disruptive force in this and other disciplines that rely on visual pattern recognition. Transfer learning using convolutional neural networks represents the state-of-the-art in artificial melanoma diagnosis, achieving similar sensitivity and specificity to experts. (1) Several machine learning algorithms, from support vector machines to neural networks, have also been applied to decision support which can guide biopsy decision-making by general practitioners and inexperienced clinicians. (2)

This presentation outlines the spectrum of artificial intelligence technologies which have shown promise for melanoma diagnosis and decision support, including recent results using deep learning to predict the need for lesion excision. This is followed by a review of current challenges facing the use of machine learning in clinical practice. Inherent difficulties include shortcomings in training datasets, risks of overfitting and poor inductive biases. There are also broader issues with model interpretability and an uncertain regulatory environment which must be resolved. (3)

Journal Title
Conference Title

Australasian Journal of Dermatology

Book Title
Edition
Volume

62

Issue

S1

Thesis Type
Degree Program
School
DOI
Patent number
Funder(s)
Grant identifier(s)
Rights Statement
Rights Statement
Item Access Status
Note
Access the data
Related item(s)
Subject

Clinical sciences

Science & Technology

Life Sciences & Biomedicine

Dermatology

Persistent link to this record
Citation

Byth, LA; Byth, JL, Artificial intelligence for melanoma diagnosis: from support vector machines to deep learning and beyond, Australasian Journal of Dermatology, 2021, 62 (S1), pp. 33-33