Characterization of Pancreas in Diabetic Patients in CT Images using Texture Analysis

Abstract

This study conducted to establish a population database for pancreas length and width using (CT) scan and classify the texture of computed tomography images of pancreas in diabetic patient and find the values of various parameter of texture prosperities for classification. The data were collected from 213 non- diabetic with no history of pancreas disease who had undergone abdominal CT scan between 2015 and 2017, and ages of 2 to 97years old. The study revealed that the pancreas measurement was 29.94 ± 6.36 for head length, 25.07 ± 5.62 mm for head AP diameter, 61.43 ± 15.36 mm for body length, 22.74 ± 6.08 mm for body AP diameter, 33.94± 9.11 mm for tail length, 19.83 ± 6.43 mm for tail AP diameter and pancreas CT number which was evaluated as Hounsfield was 49.11 ± 8.81for pancreas , 49.46 ± 8.03for spleen and 37.61 ± 5.65 for vertebral body diameter. Data were presented as mean and standard deviation (SD) for all of the variables. Showed results which were significant at P < 0.05. We measured pancreas length and width in Pancreas measured values were computed from the contour of the pancreas on each CT image. In addition to total pancreas measurements, the density was determined by CT Hounsfield (HU). And in childhood and adolescence, the pancreas measurements and CT(HU) increased linearly with age and then declines thereafter. We provide enduring population highlighting data for pancreatic parenchymal measurements in Sudanese as well pancreatic CT (HU). Also this study concern to characterize the pancreas area to head, body and tail using Gray Level Run Length Matrix (GLRLM) and extract classification features from CT images. The GLRLM techniques included eleven’s features. To find the gray level distribution in CT images it complements the GLRLM features extracted from CT images with runs of gray level in pixels and estimate the size distribution of the sub patterns. analyzing the image with Interactive Data Language software III to measure the grey level distribution of images. The results showed that the Gray Level Run Length Matrix and features give classification accuracy of pancreas head 89.2%, body 93.6 and the tail classification accuracy 93.5%. The overall classification accuracy of pancreas area 92.0%. Also this study c the calcification accuracy of the normal pancreas 100%, diabetic pancreas is 100% and over all accuracy of pancreas area100%. These relationship patients are stored in a Texture Dictionary that can be later used to automatically annotate new CT images with the appropriate pancreas area nam