Scientific literature shows that the use of Artificial Intelligence (AI) can contribute to better improve healthcare performance and quality of care. The current COVID-19 (caused by the virus SARS-CoV-2) pandemic has increased the need for, and opportunities to use AI within healthcare. Specialists on the subject believe that technologies that incorporate AI will transform the provision of health care in the next decade, and it is important to know their applications and challenges to be faced by professionals and users.

In 2019, researchers from the University of Brasília developed a project called “Artificial Neural Network Protocol in Dermatology”, aiming to validate a Convolutional Neural Network (CNN) for diagnosis purposes in an outpatient healthcare setting in Brazil. 
The aim is to use this CNN as an alternative preliminary diagnostic (screening) system, based on photographic images rather than biopsy (an invasive and risky procedure), either face-to-face or by telemedicine.

In this specific project, the researchers tried to answer the following questions:

  1. Can AI contribute to improve efficiency in screening, by reducing diagnostic time with a similar diagnostic accuracy, of pigmented lesions of skin cancer in the Brazilian population?
  2. Will AI be of importance for Brazilian general practitioners and dermatologists to make faster and more accurate screenings without the need to use invasive methods, such as biopsy?

Due to the interdisciplinary nature of the project, a vast demographic of researchers was involved, including different disciplines and levels of seniority. The CNN is being trained to screen and discriminate skin pigmented cancer lesions from other associated benign diseases, using a set of dermatological photographic images from patients. Therefore, in order to construct a robust database to train this CNN, the researchers involved in this project, are building strong partnerships with Brazilian hospitals that have a large set of dermatological images.

It is expected that the use of CNN can also be expanded as a non-invasive technology, as it is a methodology that can be embedded in cell phones and smartphones; allowing preliminary diagnosis to be made from photographic images, obtained at the time of screening in an outpatient health setting in Brazil. The ultimate impact is to reduce risks and costs, improve access to healthcare in remote areas and the integration of information into the Unified Health System database. In addition, this technology can be applied to other health conditions or management of the healthcare system health.



Banco de dados Públicos: desafios e abordagens para determinar uma estratégia de AI

Um modelo matemático para reconstrução de forma 3D a partir de imagens 2D de lesões cutâ

Técnica clínica e desafios no diagnóstico de lesões cutâneas usando AI

Redes neurais artificiais podem nos auxiliar a detectar e classificar AVC com base em imagens de tomografia computadorizada do cérebro?

Perspectivas e desafios do uso de AI em Enfermagem

AI nos cuidados da saúde: no Brasil e no mundo

Desafios e vulnerabilidades éticas e legais num mundo de AI, BIG-DATAS e Rede Neurais

Minicurso: Metodologias de AI/Redes Neurais em Dermatologia

Impactos e perigos da inteligência artificial para os sistemas de Serviços Sociais e de saúde

Aplicação de algoritmos de inteligência artificial para auxiliar no tratamento de tumores hepáticos