Method and device for extracting quantitative information in medical ultrasound

1 . A method for extracting quantitative information by a device operated by at least one processor, the method comprising:

receiving pulse-echo data obtained from sensors of an ultrasound probe according to beam patterns radiated into a tissue;

obtaining a region of interest (ROI); and

extracting quantitative information on the ROI from the pulse-echo data by using a neural network trained to extract quantitative information from input data,

wherein the neural network normalizes a feature of each sensor by using normalization parameters of each sensor extracted according to a location of the ROI.

2 . The method of claim 1 , wherein the obtaining the ROI includes

obtaining the location of the ROI in a B-mode image generated from the pulse-echo data.

3 . The method of claim 1 , further comprising:

extracting pulse-echo data of the ROI from the pulse-echo data according to a location of the ROI; and

inputting the pulse-echo data of the ROI into the neural network.

4 . The method of claim 1 , wherein the neural network is configured to

encode a quantitative feature included in the input pulse-echo data to generate an encoding profile, and

extract quantitative information from the encoding profile while normalizing the feature of each sensor through a ROI adaptive normalization layer.

5 . The method of claim 1 , wherein the quantitative information is an attenuation coefficient.

6 . The method of claim 5 , wherein the pulse-echo data is abdominal ultrasound data, and

the ROI includes a partial region of a liver.

7 . The method of claim 6 , further comprising:

extracting information on hepatic steatosis in the ROI based on the attenuation coefficient extracted from the pulse-echo data.

8 . A method for extracting quantitative information by a device operated by at least one processor, the method comprising:

receiving pulse-echo data obtained from sensors of an ultrasound probe according to beam patterns radiated to an abdomen;

providing a B-mode image generated from the pulse-echo data to an interface screen, and obtaining a region of interest (ROI) from the interface screen;

inputting the pulse-echo data and a location of the ROI to a neural network trained to extract quantitative information from input data, and extracting quantitative information on the ROI; and

providing the quantitative information and/or analysis extracted based on the quantitative information to the interface screen,

wherein the neural network includes an ROI adaptive normalization layer that normalizes a feature of each sensor by using normalization parameters of each sensor extracted based on the location of ROI.

9 . The method of claim 8 , further comprising:

extracting information on hepatic steatosis in the ROI based on the quantitative information including an attenuation coefficient.

10 . The method of claim 8 , wherein the location of the ROI is represented as a vector representing a depth and a steering angle from the ultrasound probe.

11 . The method of claim 8 , wherein the neural network is configured to

encode a quantitative feature included in the input pulse-echo data to generate n encoding profile, and

extract an attenuation coefficient of the interest region from the encoding profile while normalizing the feature of each sensor through the ROI adaptive normalization layer.

12 . A method for extracting quantitative information by a device operated by at least one processor, the method comprising:

extracting pulse-echo data of a region of interest (ROI) from pulse-echo data obtained according to beam patterns from sensors of an ultrasound probe; and

by using a neural network:

encoding the pulse-echo data of the ROI for each channel corresponding to a steering angle of a beam pattern, and integrating outputs encoded for each channel to generate an encoding profile;

normalizing a feature of each sensor included in the encoding profile by using normalization parameters of each sensor extracted based on a location of the ROI; and

extracting quantitative information on the ROI from a feature normalized adaptively to the location of the ROI.

13 . The method of claim 12 , wherein the normalization parameters of each sensor are parameters that scale and shift a feature of each sensor.

14 . The method of claim 12 , wherein the location of the ROI is represented as a vector representing a depth and a steering angle from the ultrasound probe.

15 . The method of claim 12 , wherein the extracting the quantitative information on the ROI includes

extracting the quantitative information on the ROI by using a regression network trained to extract quantitative information included in the encoding profile through a sequential regression layer.

16 . The method of claim 12 , wherein the quantitative information is an attenuation coefficient.

17 . The method of claim 16 , wherein the pulse-echo data is abdominal ultrasound data, and

the ROI includes a partial region of a liver.

18 . The method of claim 17 , further comprising:

extracting information on hepatic steatosis in the region of interest based on the attenuation coefficient of the region of interest.