Level Based Anomaly Detection of Brain MR Images Using Modified Local Binary Pattern

Abstract

The medical imaging technology plays a crucial role in visualization and analysis of the human body with unprecedented accuracy and resolution. Analyzing the multimodal for disease-specific information across patients can reveal important similarities between patients, hence their underlying diseases and potential treatments. Classification of MR brain images as normal or abnormal with information about the level at which it lies is a very important task for further processing, which is helpful for the diagnosis of diseases. This paper focuses on the abnormality detection of brain MR images using search and retrieval technique performed on similar anatomical structure images. Similar anatomical structure images are retrieved using Modified Local Binary Pattern (MOD-LBP) features of the query and target images and the level of the image is identified. The query image is compare with images in the same level and classification is done using the SVM classifier. The result reveals that the classification accuracy is improved significantly when the query image is compared with similar anatomical structure images.