Brain tumor MRI medical images classification with data augmentation by transfer learning of VGG16

• Main Authors: Ahmed Yahya Al-Galal, Sabaa, Taha Alshaikhli, Imad Fakhri, Abdulrazzaq, M. M., Hassan, Raini
• Format: Article
• Language: English; English
• Published: School of Engineering, Taylor’s University 2021
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://irep.iium.edu.my/95278/1/ACSAT%202021%20%28Sabaa%20et%20al.%29.pdf; http://irep.iium.edu.my/95278/2/ACSAT%202021%20%28Sabaa%20et%20al.%29.pdf; http://irep.iium.edu.my/95278/; https://jestec.taylors.edu.my/
• Institution: Universiti Islam Antarabangsa Malaysia

The ability to estimate conclusions without direct human input in healthcare systems via computer algorithms is known as Artificial intelligence (AI) in healthcare. Deep learning (DL) approaches are already being employed or exploited for healthcare purposes, and in the case of medical images analysis, DL paradigms opened a world of opportunities. This paper describes creating a DL model based on transfer learning of VGG16 that can correctly classify MRI images as either (tumorous) or (non-tumorous). In addition, the model employed data augmentation in order to balance the dataset and increase the number of images. The dataset comes from the brain tumour classification project, which contains publicly available tumorous and non-tumorous images. The result showed that the model performed better with the augmented dataset, with its validation accuracy reaching ~100 %.