patologIA: Online platform for improving image-based diagnostics applied to medical students

Autor(es)

Lima, Cesar A. S. ; dos Santos, Clayton Eduardo ; Bissaco, Marcia A. S. ; da Silva, Alessandro P. ; Scardovelli, Terigi A. ; Boschi, Silvia R. M. da S. ; Martini Rodrigues, Silvia Cristina ; Deserno, TM ; Park, BJ

• Tipo de documento: Artigo Científico
• Fonte: MEDICAL IMAGING 2021: IMAGING INFORMATICS FOR HEALTHCARE, RESEARCH, AND APPLICATIONS; v. 11601, p. 15-pg., 2021-01-01.

Resumo

The present work consists of the development of an online computational platform used in the classroom or not as a learning aid tool, in particular, for the interpretation of medical images with diagnosis based on images, such as x-ray, tomography, resonance and the like. The resources offered will allow the student to identify patterns of pathologies in different regions of the human body with different levels of difficulty. Each image will be classified on a progressive scale from 1 to 4, according to the degree of difficulty and according to the specifics of the report, leading the student to an increasing level of learning. The tool also offers forms with questions related to the clinical cases presented, prepared by the teacher, which must be answered by the student after analyzing each case. The platform is modular, so it was designed to receive new data and functionality constantly. In this sense, new images, without reports, when inserted will be classified using artificial intelligence techniques based on deep learning, such as CNN's - Convolutional Neural Networks. The application development is strongly supported in the cloud computing paradigm and its APIs - Application Programming Interface, as well as in the python programming language and in open source libraries, such as Detectron2 and PyTorch, necessary for the implementation of image processing resources aimed at object detection. For data formatting of each type of image, the standard chosen was COCO - Common Objects in Context. (AU)

• Processo FAPESP: 17/14016-7 - Ambiente web para aprendizado e treinamento de graduandos, residentes ou médicos na interpretação de imagens de mama
• Beneficiário: Silvia Cristina Martini Rodrigues
• Modalidade de apoio: Auxílio à Pesquisa - Regular