Characterization of Diffuse Liver Diseases using Ultrasonography and Texture Analysis

Abstract

This is a descriptive study which was carried out during the period from September-2013 to May 2016 in (Ibn Sena, Ribat and college of Medical Radiologic Sciences) Khartoum- Sudan. This study discusses the depiction of diffuse liver diseases using ultrasonography and texture analysis. A total of “200” patients were selected randomly; all patients have ages above 20 years, they include: patient with normal liver, hepatitis B, hepatitis C and liver cirrhosis; all referred by physician. Any patient with other liver disorder was excluded from this study. All patients were examined by ultrasound scanning using ‘Honda’ Eucup and General Electric scanners with 3,5MHz probe. Trans abdominal Scanning were performed for all patients to evaluate liver shape; i.e. if it is normal or abnormal; also evaluate liver texture, left edge of the liver, the liver span measure, right lobe size, caudate lobe size, portal vein diameter, presence of ascites and spleen size. Data was collected using master data collection sheet where it was analyzed using crosstabulation, linear regression, discriminant analysis and pattern recognition. The results of this study shows that the normal liver can be identified by an accuracy of (83.3%) while the sensitivity of diagnosing liver cirrhosis was 98.3%, for hepatitis type B the sensitivity was 91.7% where 8.3% of the cases showed measures similar to the normal liver; this occurs specially in early stage. But for hepatitis type C the sensitivity was 45% where 55% of the cases showed measures similar to the normal liver. Study shows that; the U/S findings of the liver cirrhosis mostly changes shape 52 patients (87%) out of 60 cases with liver cirrhosis, changed texture (hyper echoic and nodular) 55 patients (95%), (Abnormal shape or abnormal texture and increased the C/R ratio, presence of ascites and splenomegaly). Study revealed that the ultrasound scanning alone is not reliable and accurate in diagnosis of hepatitis C, but reliable and accurate in differentiate between normal liver and disorder liver. In addition to that the study concluded that: Computer aided texture analysis is very Combination of ultrasound scanning with texture analysis increased the accuracy of diffuse liver classification. Study recommended that: If there is a prediction of liver disorder by U/S scanning, the patient images must be analyze using IDL program, then the histopathology will be done for confirmation only, if it is not available, no need of it for giving the patient treatment, because the accuracy of U/S and texture analysis tools together is very high