Brain tumor X-ray images enhancement and classification using anisotropic diffusion filter and transfer learning models

• Published in: International journal of information technology (Singapore. Online) 2024, Vol.16 (6), p.3771-3779
• Main Authors: Gomaa, Mamdouh M., Zain elabdeen, Asmaa G., Elnashar, Alaa, Zaki, Alaa M.
• Format: Article
• Language: English
• Subjects: Artificial Intelligence; Computer Imaging; Computer Science; Image Processing and Computer Vision; Machine Learning; Original Research; Pattern Recognition and Graphics; Software Engineering; Vision
• ISSN: 2511-2104; 2511-2112
• DOI: 10.1007/s41870-024-01830-0

One of the diseases with the fastest rate of spread is brain tumors, which affect millions of people. Thus, brain tumor classification is intensively studied. In general, several imaging techniques, such as magnetic resonance imaging (MRI), computed tomography (CT), and x-radiation (X-ray), have been used for assessing brain tumors. Recent advancements in deep learning have aided the medical imaging sector in the diagnosis of numerous ailments. In this paper, we used a suitable filter for noise removal so that the image's edge is preserved without losing information that is essential in the interpretation of the brain tumor images, then data augmentation to generate variations of the dataset to provide artificial data for training, and finally deep learning to classify brain x-ray images as tumor or non-tumor. Three pretrained fine-tuning deep learning models, Visual Geometry Group 19 (VGG19), Inception V3, and Mobile Network (MobileNet V2), are used to classify brain tumors. According to the results, VGG19 is the better model, with an accuracy of 98.58% compared to InceptionV3 and MobileNetV2, which have accuracy values of 97.6% and 98.47%, respectively.