Saliency-Based Brain Abnormality Identification in Magnetic Resonance Images

Abstract

Due to the great development in the field of digital image processing, it has become an integral and important part of computer-aided diagnosis systems. Magnetic resonance image (MRI) processing is one of the hot fields that attracted researchers due to its role in improving and speeding up diagnosis. This work presents a new method for identifying abnormal regions in brain MR images using saliency recognition techniques. This is because tumour regions or abnormal regions share with saliency the fact that these regions differ from the rest of the image in terms of size, luminance, and texture. In this paper, we use a saliency extraction method that uses the irregularity of the regions in its definition, as this method depends on the basis that the salient areas in the image are rare and different from the rest of the image, which is what we need to determine the tumour area. The proposed algorithm is applied to a standard dataset and the obtained result showed a high level of accuracy, where the precision, recall, F-measure, and accuracy averages reached 91.8%, 96.2%, 84.5%, and 96% respectively. The results were discussed thoroughly, and the limitations were identified and discussed as well.

Symmetry-Based Brain Abnormality Detection Using Machine Learning

“…where (𝐿 , 𝑅 ) is an ordered pair that contains the left 𝐿 and the right 𝑅 halves of the image 𝒥 . Each half is then undergone a metrics extraction process which can be done using the mapping (ℱ) given in equation (3).…”

“…In literature, MR image processing is divided into five stages, namely, pre-processing, feature extraction, feature selection [2], classification, and segmentation [3], [4]. The pre-processing stage includes processes such as filtering, enhancement and registration.…”