Machine Learning in Melanoma Diagnosis. Limitations About to be Overcome.
Uso del aprendizaje automático en el diagnóstico del melanoma. Limitaciones por superar.
Aprendizaje automático
Artificial Intelligence
Clasificación de imágenes
Convolutional neural networks
Cáncer de piel
Dermatoscopia
Dermoscopy
Image classification
Inteligencia artificial
Machine learning
Melanoma
Redes neuronales convolucionales
Skin cancer
Journal
Actas dermo-sifiliograficas
ISSN: 2173-5778
Titre abrégé: Actas Dermosifiliogr (Engl Ed)
Pays: Spain
ID NLM: 101777537
Informations de publication
Date de publication:
May 2020
May 2020
Historique:
received:
11
08
2019
accepted:
16
09
2019
pubmed:
7
4
2020
medline:
25
6
2021
entrez:
7
4
2020
Statut:
ppublish
Résumé
Automated image classification is a promising branch of machine learning (ML) useful for skin cancer diagnosis, but little has been determined about its limitations for general usability in current clinical practice. To determine limitations in the selection of skin cancer images for ML analysis, particularly in melanoma. Retrospective cohort study design, including 2,849 consecutive high-quality dermoscopy images of skin tumors from 2010 to 2014, for evaluation by a ML system. Each dermoscopy image was assorted according to its eligibility for ML analysis. Of the 2,849 images chosen from our database, 968 (34%) met the inclusion criteria for analysis by the ML system. Only 64.7% of nevi and 36.6% of melanoma met the inclusion criteria. Of the 528 melanomas, 335 (63.4%) were excluded. An absence of normal surrounding skin (40.5% of all melanomas from our database) and absence of pigmentation (14.2%) were the most common reasons for exclusion from ML analysis. Only 36.6% of our melanomas were admissible for analysis by state-of-the-art ML systems. We conclude that future ML systems should be trained on larger datasets which include relevant non-ideal images from lesions evaluated in real clinical practice. Fortunately, many of these limitations are being overcome by the scientific community as recent works show.
Sections du résumé
BACKGROUND
BACKGROUND
Automated image classification is a promising branch of machine learning (ML) useful for skin cancer diagnosis, but little has been determined about its limitations for general usability in current clinical practice.
OBJECTIVE
OBJECTIVE
To determine limitations in the selection of skin cancer images for ML analysis, particularly in melanoma.
METHODS
METHODS
Retrospective cohort study design, including 2,849 consecutive high-quality dermoscopy images of skin tumors from 2010 to 2014, for evaluation by a ML system. Each dermoscopy image was assorted according to its eligibility for ML analysis.
RESULTS
RESULTS
Of the 2,849 images chosen from our database, 968 (34%) met the inclusion criteria for analysis by the ML system. Only 64.7% of nevi and 36.6% of melanoma met the inclusion criteria. Of the 528 melanomas, 335 (63.4%) were excluded. An absence of normal surrounding skin (40.5% of all melanomas from our database) and absence of pigmentation (14.2%) were the most common reasons for exclusion from ML analysis.
DISCUSSION
CONCLUSIONS
Only 36.6% of our melanomas were admissible for analysis by state-of-the-art ML systems. We conclude that future ML systems should be trained on larger datasets which include relevant non-ideal images from lesions evaluated in real clinical practice. Fortunately, many of these limitations are being overcome by the scientific community as recent works show.
Identifiants
pubmed: 32248945
pii: S0001-7310(20)30004-1
doi: 10.1016/j.ad.2019.09.002
pii:
doi:
Types de publication
Journal Article
Langues
eng
spa
Sous-ensembles de citation
IM
Pagination
313-316Informations de copyright
Copyright © 2020 AEDV. Publicado por Elsevier España, S.L.U. All rights reserved.