Analysis of Kidney Ultrasound Images Characterization Using Statistical Moment Descriptor

Abstract

Commonly the abnormalities of renal diseases were diagnosed and described using morphological or radiological nomenclature. In distinguishing abnormalities, especially kidney diseases, it is important to characterize the ultrasound images more objectively. This study used The Statistical Moment Descriptor (SMD) with the pixel-level spatial distribution of B-mode ultrasound to analyze three types of the area such as the full kidney, the renal pelvis, and the cortex of fifty kidney ultrasound images. The significant SMD properties that can distinguish the normal from the abnormal ones with 95% confident level are the Mean (p=1.11e-04), the Median (p=0.0051), the Kurtosis (p=0.0053), the Standard Deviation (p=0.0082), and the Entropy (p=0.019) unlike Range (p= 0.091) and Skewness (p=0.1389). Renal pelvis and cortex areas plotting method is unable to distinguish the abnormalities for the CKD on some properties.