ABSTRACT
The polymorphism of cutaneous leishmaniasis (CL) complicates diagnosis in health care services because lesions may be confused with other dermatoses such as sporotrichosis, paracocidiocomycosis, and venous insufficiency. Automated identification of skin diseases based on deep learning (DL) has been applied to assist diagnosis. In this study, we evaluated the performance of AlexNet, a DL algorithm, to identify pictures of CL lesions in patients from Midwest Brazil. We used a set of 2458 pictures (up to 10 of each lesion) obtained from patients treated between 2015 and 2022 in the Leishmaniasis Clinic at the University Hospital of Brasilia. We divided the picture database into training (80%), internal validation (10%), and testing sets (10%), and trained and tested AlexNet to identify pictures of CL lesions. We performed three simulations and trained AlexNet to differentiate CL from 26 other dermatoses (e.g., chromomycosis, ecthyma, venous insufficiency). We obtained an average accuracy of 95.04% (Confidence Interval 95%: 93.81-96.04), indicating an excellent performance of AlexNet in identifying pictures of CL lesions. We conclude that automated CL identification using AlexNet has the potential to assist clinicians in diagnosing skin lesions. These results contribute to the development of a mobile application to assist in the diagnosis of CL in health care services.
ABSTRACT
RESUMO Deep Learning é uma técnica de aprendizado de máquina na qual o programa computacional aprende padrões diretamente a partir de imagens classificadas previamente. O presente ensaio objetivou apresentar essa técnica e algumas de suas aplicações para diagnóstico de doenças e identificação de insetos vetores para incentivar profissionais da saúde que não tenham conhecimento aprofundado em informática e que desejem utilizar a ferramenta para realizar análises automatizadas. Deep Learning tem sido aplicado para diagnóstico de câncer, fibrose cardíaca, tuberculose, detecção de parasitos como Plasmodium e Leishmania e ainda para identificação de insetos vetores. Na Universidade de Brasília, a técnica tem sido aplicada para desenvolver uma ferramenta para identificar lesões ulceradas de leishmaniose em diagnóstico diferencial e para detectar Leishmania em lâminas de estudos histopatológicos. Além disso, Deep Learning tem sido usado para identificar as espécies de vetores da doença de Chagas - o que é importante para auxiliar na vigilância epidemiológica. O uso da tecnologia envolve desafios éticos e procedimentais que são discutidos no presente ensaio. O ensaio aponta perspectivas de desenvolvimento de aplicativos que auxiliem os profissionais de saúde no diagnóstico de Leishmaniose e de vetores da doença de Chagas, o que vai ao encontro dos objetivos da pesquisa translacional.
ABSTRACT Deep Learning is a machine learning technique in which the computational algorithm learns patterns directly from images previously classified. The present essay aims to show some of its applications for clinical diagnosis and identification of insect vectors to encourage health professionals who do not have deep knowledge of computer science and who wish to use the tool to perform automated analyzes. Deep Learning has been applied to the diagnosis of cancer, cardiac fibrosis, tuberculosis, detection of parasites such as Plasmodium and Leishmania, and to identify insect vectors. At the University of Brasília, Deep Learning has been used to develop a tool to identify ulcers caused by leishmaniasis, as well as to detect Leishmania parasites. Moreover, Deep Learning was applied to identify the species of vectors of Chagas disease, an important contribution to the epidemiological surveillance of the disease. The use of Deep Learning involves some ethical and procedural issues that are discussed in this paper. Finally, the essay points out perspectives of development of apps that assist health professionals in the diagnosis of Leishmaniasis and Chagas disease vectors, which meets the goals of translational research.
