Method and device for quantitative imaging in medical ultrasound

1 . A method of operating an imaging device, wherein the method is performed by at least one processor, the method comprising:

receiving pulse-echo data obtained from a tissue;

generating a B-mode image from the pulse-echo data;

in response to a selection of an objective variable, extracting a style feature of the objective variable from the pulse-echo data;

generating style-transferred content features by applying the style feature on content features of the B-mode image; and

reconstructing a quantitative image of the objective variable using the style-transferred content features,

wherein the style feature varies according to the selected objective variable.

2 . The method of claim 1 , wherein the pulse-echo data is radio frequency (RF) data in which ultrasound signals incident on the tissue with different beam patterns are reflected on the tissue, and returned.

3 . The method of claim 1 , wherein the objective variable includes at least one of Speed of Sound (SoS), Attenuation Coefficient (AC), Effective Scatterer Concentration (ESC), or effective scatterer diameter (ESD).

4 . The method of claim 1 , wherein the generating style-transferred content features includes

style-transferring the content features of the B-mode image using the style feature to generate the style-transferred contents in form of images.

5 . The method of claim 1 , wherein the extracting the style feature of the objective variable includes

extracting the style feature of the selected objective variable from the pulse-echo data, through conditional encoding according to the selected objective variable.

6 . The method of claim 1 , wherein the content features of the B-mode image include geometric information of the tissue in the B-mode image.

7 . The method of claim 1 , wherein the content features of the B-mode image include multi-resolution content features.

8 . The method of claim 1 , wherein the reconstructing the quantitative image of the objective variable includes

generating a high-resolution quantitative image by using parallel multi-resolution subnetworks.

9 . The method of claim 1 , wherein the extracting the style feature of the objective variable includes:

selecting normalization parameters according to the objective variable; and

extracting the style feature of the objective variable from the pulse-echo data by performing normalization using the normalization parameters,

wherein the normalization parameters include a scaling value and a shift value that are selected from a matrix according to the objective variable, and

wherein the performing normalization using the normalization parameters includes performing conditional instance normalization (CIN) using the scaling value and the shift value.

10 . The method of claim 1 , wherein the generating style-transferred content features includes

performing spatially adaptive demodulation by spatially adaptively transferring each of the content features having a size of W×H into an image having C channels through the styling feature expressed with a spatial resolution R W×H×C , where W, H, and C are integers.

11 . A method of operating an imaging device, wherein the method is performed by at least one processor, the method comprising:

in response to a selection of an objective variable, extracting a style feature of the objective variable included in pulse-echo data by using the style encoder;

generating a B-mode image from the pulse-echo data, and extracting content features of the B-mode image, by using a B-mode encoder; and

reconstructing a quantitative image of the objective variable by applying the style feature of the objective variable on content features of the B-mode image, by using a decoder,

wherein the style encoder is a neural network trained to extract a style feature of a selected objective variable from the pulse-echo data through conditional encoding according to the selected objective variable, and

wherein the style feature varies according to the selected objective variable.

12 . The method of claim 11 , wherein the B-mode encoder is a neural network trained to generate a B-mode image from the pulse-echo data, and extract the content features from the B-mode image, and

wherein the content features include geometric information.

13 . The method of claim 11 , wherein the decoder is a neural network trained to receive content features extracted from the B-mode encoder, style-transfer the content features of the B-mode image into quantitative images by using the style feature, and reconstruct the quantitative image of the objective variable using the style-transferred content features.

14 . The method of claim 13 , wherein the decoder is a neural network with a structure of parallel multi-resolution subnetworks.

15 . The method of claim 11 , wherein the objective variable includes at least one of Speed of Sound (SoS), Attenuation Coefficient (AC), Effective Scatterer Concentration (ESC), or effective scatterer diameter (ESD).

16 . An imaging device comprising:

a memory configured to store instructions; and

a processor configured to execute the instructions to:

receive pulse-echo data obtained from a tissue;

generate a B-mode image from the pulse-echo data;

in response to a selection of an objective variable, extract a style feature of the objective variable from the pulse-echo data;

generate style-transferred content features by applying the style feature on content features of the B-mode image; and

reconstruct a quantitative image of the objective variable using the style-transferred content features,

wherein the style feature varies according to the selected objective variable.

17 . The imaging device of claim 16 , wherein the processor is configured to

style-transfer the content features of the B-mode image using the style feature to generate the style-transferred contents in form of images.

18 . The imaging device of claim 16 , wherein the processor is configured to

extract the style feature of the selected objective variable from the pulse-echo data, through conditional encoding according to the selected objective variable.

19 . The imaging device of claim 16 , wherein the objective variable includes at least one of Speed of Sound (SoS), Attenuation Coefficient (AC), Effective Scatterer Concentration (ESC), or effective scatterer diameter (ESD).