From a heap of brain CT images, manual search to identify disease effected abnormal candidates is a hectic task and time-consuming. The artificial intelligence and fast computing capacity of the computer are used to design a system which will identify images with abnormality due to the effect of brain diseases from a large dataset automatically. The complete dataset will be divided into two categories-normal and abnormal. A filtering process is followed by texture analysis and clustering to conclude a slice's category. Entire image dataset is passed through a band reject filter to remove the normal area of brain before feature extraction. Binary information descriptor and second order texture features are used for classification. Significant improvement in classification accuracy is achieved by using this filtering process presented in this article. © 2019 IEEE.