Detecção de tumor cerebral a partir de análise de imagens médicas usando inteligência artificial

Abstract

The use of medical imaging is evident in several hospital areas. This work brings special attention to magnetic resonance imaging in patients with healthy brains or with tumors. Therefore, it is interesting to make a connection between the medical and the computational areas in order to meet the need to study more efficient approaches in relation to image quality, execution time, analysis of results and patient beneĄt. One of the most studied areas currently in the world of computing is ArtiĄcial Intelligence (AI). Applying in this work it will be used as a tool for detecting tumors from pre-processed images and if it returns a positive result, be able to locate each region classiĄed as a brain tumor. For this, a convolutional neural network was created using the Python programming language with the help of some libraries, such as Pytorch, and a technique called data augmentation. Such a network was taught from the set of medical images gathered in a dataset available on the Figshare website (CHENG, 2017). With this, satisfactory results were obtained with accuracy of 98%, where the model generated by the network was able to Ąnd images with tumors and point out their location.