Diabetic retinopathy classification for supervised machine learning algorithms

International Journal of Retina and Vitreous

Luis Filipe Nakayama¹  Fernando Korn Malerbi¹  Helen Nazareth Veloso dos Santos¹  Caio Vinicius Saito Regatieri¹  Lucas Zago Ribeiro¹  Mauricio Maia¹  Paulo Henrique Morales²  Rubens Belfort Mattos²  Mariana Batista Gonçalves³  Daniel A. Ferraz³ 

[1]Physician, Department of Ophthalmology, Universidade Federal de São Paulo - EPM, Botucatu Street, 821, Vila Clementino, 04023-062, São Paulo, SP, Brazil

[2]Physician, Department of Ophthalmology, Universidade Federal de São Paulo - EPM, Botucatu Street, 821, Vila Clementino, 04023-062, São Paulo, SP, Brazil

[3]Instituto Paulista de Estudos e Pesquisas em Oftalmologia, IPEPO, Vision Institute, São Paulo, SP, Brazil

[4]Physician, Department of Ophthalmology, Universidade Federal de São Paulo - EPM, Botucatu Street, 821, Vila Clementino, 04023-062, São Paulo, SP, Brazil

[5]Instituto Paulista de Estudos e Pesquisas em Oftalmologia, IPEPO, Vision Institute, São Paulo, SP, Brazil

[6]NIHR Biomedical Research Centre for Ophthalmology, Moorfield Eye Hospital, NHS Foundation Trust, and UCL Institute of Ophthalmology, London, UK

关键词: Diabetic retinopathy classifications;    Artificial intelligence;    Datasets;   

DOI  :  10.1186/s40942-021-00352-2

来源: Springer

【 摘 要 】

BackgroundArtificial intelligence and automated technology were first reported more than 70 years ago and nowadays provide unprecedented diagnostic accuracy, screening capacity, risk stratification, and workflow optimization.Diabetic retinopathy is an important cause of preventable blindness worldwide, and artificial intelligence technology provides precocious diagnosis, monitoring, and guide treatment. High-quality exams are fundamental in supervised artificial intelligence algorithms, but the lack of ground truth standards in retinal exams datasets is a problem.Main bodyIn this article, ETDRS, NHS, ICDR, SDGS diabetic retinopathy grading, and manual annotation are described and compared in publicly available datasets. The various DR labeling systems generate a fundamental problem for AI datasets. Possible solutions are standardization of DR classification and direct retinal-finding identifications.ConclusionReliable labeling methods also need to be considered in datasets with more trustworthy labeling.

【 授权许可】

CC BY