AI-based system for detecting and locating brain tumors on cranial magnetic resonance imaging

Authors

Landazuri Galarza, Edison Sebastian

Changoluiza Cundulli, Jefferson Xavier

Tapia, Freddy

Tello Oquendo, Luis

Abstract

Brain tumors or growths of abnormal cells in the brain tissue is a problem that is becoming more and more common in the population. Generally, the most common symptoms are headaches, personality changes (depression, anxiety, or disinhibited), loss of balance, and difficulty concentrating. The developed prototype serves as a preventive instrument to detect possible anomalies associated with these tumors; a fusion between Artificial Intelligence (AI) and Neuroimaging was devised. Besides using some images contained in datasets (3 339 images), some training and validation processes were conducted through Google Colab, where the formats of the images are essential when evidencing some anomaly. Furthermore, using pretrained neural networks with the deep search networks facilitated the training of the AI-based model and increased the accuracy of the results.