A Novel Statistical Approach to Obtain the Best Visibility Slice in MRI Sequence of Brain Tumors

Abstract

Early diagnosis of brain tumor enhances the possibility of patients being cured. With the progress of the use of artificial intelligence in the medical field, the detection of brain tumors has become one of the researchers’ interests. Obtaining which slice in the MRI sequence gives the best visibility of the tumor is still a challenge. This paper introduced a novel statistical approach to extracting the tumor from the patient's MRI scan (sequence). Initially, the probability mass function (PMF) was computed for each image in the sequence. Then, the Kullback-Leibler divergence technique was applied to determine the tumor image(s) that diverged from the respective healthy ones. The best tumor visibility slice was determined using the root mean square error metric. In addition, a clustering approach was applied to segment the tumor images. The vector quantization (VQ) method was utilized for grouping the images into 16 different clusters, while a reverse VQ technique was employed to produce two-tone images. Finally, a 2D Teager operator was used to detect the edges for tumor demarcation. A private dataset of twenty MRI scans (sequences) was used for testing and evaluating the system.

Keywords

• Brain tumor, Kullback-Leibler divergence, MRI images, Probability mass function, Tumor detection