Apparatus for ultrasound diagnosis of liver steatosis using feature points of ultrasound image and remote medical-diagnosis method using the same

An apparatus for automatic ultrasound diagnosis of liver steatosis using feature points in an ultrasound image, the apparatus comprising:

• an ultrasound probe sensor acquiring an ultrasound image from a patient;
• a template kidney feature point orientation data storage unit storing orientation data with respect to template kidney periphery feature points corresponding to feature points of a kidney periphery in template ultrasound pictures and pixels in a patch image of each of the feature points;
• a template feature point statistical data storage unit storing statistical data with respect to feature points of a kidney region in the template ultrasound pictures and feature points of regions of interest excluding the kidney region;
• a feature point detection unit detecting feature points from the ultrasound image or from an edge component of the ultrasound image;
• an orientation data calculation unit calculating orientation data with respect to pixels in a patch image of each of the feature points of the ultrasound image;
• a bad feature point removal unit removing bad feature points from the feature points of the ultrasound image to locate feature points for comparison;
• a major orientation correction unit obtaining the most frequent orientation as a major rotational orientation in a histogram indicating the frequency of pixels with respect to an orientation, based on orientation data with respect to pixels in a patch image of each of the feature points for comparison, and correcting orientations of the pixels in the patch image of each of the feature points for comparison based on the major rotational orientation
• a feature point-pair generation unit generating a pair of feature points between the template kidney periphery feature points and the feature points for comparison pertaining to patch images having high similarity through comparison between the orientation data with respect to the pixels in the patch images of the template kidney periphery feature points and in the patch images of the feature points for comparison;
• a curve fitting unit generating a first fitting curve through ellipse fitting or curve fitting with respect to an imaginary closed loop formed by connecting the feature points for comparison paired with the template kidney periphery feature points by the feature point-pair generation unit, or calculating the number of fitting feature points corresponding to the number of feature points overlapping each other on two-dimensional space coordinates through scanning comparison between a template edge component and the feature points for comparison paired with the template kidney periphery feature points by the feature point-pair generation unit;
• a kidney detection unit detecting a location of a kidney from the ultrasound image based on the first fitting curve or the number of fitting feature points;
• a region-of-interest extraction unit extracting images of one or more regions of interest through space addressing on the ultrasound image with reference to the detected kidney;
• a feature point statistical data calculation unit calculating feature point statistical data from the feature points of the kidney region detected by the kidney detection unit or from the feature points of the regions of interest extracted by the region-of-interest extraction unit;
• and
• a fatty liver grade determination unit determining a grade of liver steatosis classified into a mild grade, a moderate grade, a severe grade, and a cirrhosis grade, based on the template feature point statistical data and the feature point statistical data obtained from the feature point statistical data calculation unit,
• wherein the grade of liver steatosis of the patient is automatically determined based on the feature point statistical data of the feature points obtained from the ultrasound images of the patient input from the ultrasound image sensor.

An apparatus for automatic ultrasound diagnosis of liver steatosis using feature points in an ultrasound image, the apparatus comprising:

• an ultrasound probe sensor acquiring an ultrasound image from a patient;
• a template diaphragm feature point orientation data storage unit storing orientation data with respect to template diaphragm feature points corresponding to feature points of the diaphragm included in template ultrasound pictures and pixels in a patch image of each of the feature points;
• a template diaphragm orientation displacement storage unit obtaining candidate orientations having the maximum magnitudes from each of the patch images corresponding to the template diaphragm feature points and storing, as template diaphragm orientation displacement data, orientation displacement data corresponding to a difference between the maximum orientation and the minimum orientation among the candidate orientations;
• a feature point detection unit detecting feature points in the ultrasound image;
• an orientation data calculation unit calculating orientation data with respect to pixels in a patch image of each of the feature points of the ultrasound image;
• a bad feature point removal unit removing bad feature points from the feature points obtained from the ultrasound image to locate feature points for comparison, the bad feature points being determined by considering feature points of the ultrasound image deviating from a template diaphragm orientation displacement range to be the bad feature points;
• a major orientation correction unit obtaining the most frequent orientation as a major rotational orientation in a histogram indicating the frequency of pixels with respect to an orientation, based on orientation data with respect to pixels in a patch image of each of the feature points for comparison, and correcting orientations of the pixels in the patch image of each of the feature points for comparison based on the major rotational orientation;
• a feature point-pair generation unit generating a pair of feature points between the feature points using patch images having high similarity through comparison between the orientation data with respect to the pixels in the patch images of the template diaphragm feature points and in the patch images of the feature points for comparison;
• a curve fitting unit generating a two-dimensional curve fitting curve through curve fitting with respect to an imaginary closed loop formed by connecting the feature points for comparison paired with the template diaphragm feature points by the feature point-pair generation unit;
• a diaphragm image extraction unit extracting an image of a diaphragm region from the two-dimensional curve fitting curve;
• a fitting distance calculator calculating a Euclidean distance between the two-dimensional curve fitting curve and the feature points for comparison applied to curve fitting;
• a region-of-interest image extraction unit extracting an image of a region of interest from the ultrasound image through detection of a location of the region of interest based on a location of the extracted image of the diaphragm region;
• and
• a fatty liver grade determination unit determining a grade of liver steatosis classified into a mild grade, a moderate grade, a severe grade, and a cirrhosis grade, based on a brightness ratio of the region of interest to the diaphragm region obtained by calculating an average brightness of pixels in the extracted image of the region of interest and an average brightness of pixels in the extracted image of the diaphragm region,
• wherein the grade of liver steatosis of the patient is automatically determined based on the brightness ratio of the region of interest to the diaphragm region obtained from the ultrasound image of the patient input from the ultrasound probe sensor.