Development of Artificial Intelligence algorithm for diagnostic aid for hepatic lesions

• Grant number: 17/15770-7
• Support type: Research Grants - Innovative Research in Small Business - PIPE
• Duration: August 01, 2018 - April 30, 2019
• Field of knowledge: Health Sciences - Medicine - Medical Radiology
• Principal researcher: Luis Gustavo Rocha Vianna
• Grantee: Luis Gustavo Rocha Vianna
• Company: Machiron Desenvolvimento de Sistemas Ltda
• CNAE: Desenvolvimento de programas de computador sob encomenda ; Desenvolvimento e licenciamento de programas de computador customizáveis ; Tratamento de dados, provedores de serviços de aplicação e serviços de hospedagem na internet
• City: São Paulo
• Pesquisadores principais: Ana Cláudia Martins Ciconelle
• Assoc. researchers: Suzane Kioko Ono
• Associated grant(s): 19/05723-7 - LivIA - a tool for diagnostic aid for hepatic lesions, AP.PIPE
• Associated scholarship(s): 18/14866-3 - Development of artificial intelligence algorithm for diagnostic aid for hepatic lesions, BP.TT; 18/14788-2 - Development of artificial intelligence algorithm for diagnostic aid for hepatic lesions, BP.PIPE

Abstract

There is a trend in the development of new technologies for improving medicine with recent advances in computer science, especially in artificial intelligence and machine learning. One of the reasons for this trend is that, nowadays, medicine has a bigger capacity for performing exams due to the technology evolution, reducing the time spend in the exam and optimizing the workflow in imaging services. However, the capacity and the time spend of analysis by radiologists don't grow at the same rate, causing in an increase of exams to be analyzed by each doctor. It is important to note that this is a global trend in both private and public medical services. The MaChiron Company was created in this scenery with the goal to promote computational solutions to different healthcare problems following the needs of hospitals and medical laboratories. In this project, we will develop a software to perform a pre-diagnosis in abdominal Computer Tomography (CT) images classifying the scans according to the chance of hepatic disorders, with especial attention to cases of hepatocellular carcinoma (HCC). The idea for this project came from the "Hospital das Clínicas" of the Faculty of Medicine of the University of São Paulo, where there is a growing demand for CT exams to provide early HCC detection in high-risk patients. Therefore, our algorithm will decrease the time required for the analysis of CT exams by previously indicating which have suspicious lesions so that these are given priority in the radiology workflow. The algorithm can also improve the correctness of the analysis as it gives extra information for the radiologists. Our research to develop this classifier will use techniques in automatic organ segmentation, lesion detection, texture feature extraction in different regions of interest and supervised machine learning based on those features. The main challenge of this project is to integrate all those automatic techniques into one algorithm, ensuring that the final classification will be efficient and precise. Our evaluation will be the sensitivity and specificity of the method applied in the training set. (AU)