Characterization of Brain Tumor in Magnetic Resonance Images using Patterns Recognition

Abstract

This study was carried out in order to characterization of brain tumor in MRI by applying pattern recognition to the brain tissues represented on MRI to recognizes the brain tumors from the other brain tissues which included: grey and white matter, CSF and brain tumor. This study was carried out in the period from March 2018 to March 2021in Khartoum state at Radiation and Isotopes Center of Khartoum (RICK), Aliaa Specialist Hospital and Modern Medical center. The images were obtained by Philips inters 1.5 Tesla MRI systems. The data of this study collected from 150 patients having axial, sagittal and coronal views that include brain tumor and they were selected randomly from a set of 500 patients. The data were extracted from the image using 3×3, pixels window inside the window the first order and higher order statistics were calculated and used to classify the brain MRI into one of the four tissues mentioned earlier. The window scans the whole image by interlacing it one pixel horizontally, then start again from the send line when the above one was completed till the end of the image. The results of this study showed that the overall accuracy of classification process was 95.8% and for the tumor the sensitivity was 88.3% and the specificity was 98.8. In conclusion these results showed that brain tumor can be classified successfully and delineated using texture analysis with a minimum effort.