Methods and systems for projection profile enabled computer aided detection (CAD)

1. A system, comprising:

an ultrasound device, comprising at least one processor, wherein the ultrasound device is configured to:

generate volumetric ultrasound dataset, based on echo ultrasound signals;

generate based on the volumetric ultrasound dataset, a three-dimensional (3D) ultrasound volume; and

apply selective structure detection to the three-dimensional (3D) ultrasound volume, the selective structure detection comprising:

generating based on a projection of the three-dimensional (3D) ultrasound volume in a particular spatial direction, a two-dimensional (2D) image;

applying two-dimensional (2D) structure detection to the two-dimensional (2D) image, wherein the two-dimensional (2D) structure detection is configured to identify structure candidates associated with a particular type of structures, the particular type of structures comprising abnormal tissue structures in a particular body part;

selecting for each identified structure candidate, a corresponding local volume within the three-dimensional (3D) ultrasound volume;

applying three-dimensional (3D) structure detection to each selected local volume, wherein:

the three-dimensional (3D) structure detection is configured to compare characteristics associated with the structure candidate with the particular type of structures, and

the three-dimensional (3D) structure detection is applied at full-resolution to each selected local volume; and

identifying based on applying the three-dimensional (3D) structure detection, one or more structure candidates that match the particular type of structures.

2. The system of claim 1 , wherein the ultrasound device is configured to apply the selective structure detection based on a projection of the three-dimensional (3D) ultrasound volume in at least one other spatial direction.

3. The system of claim 2 , wherein the ultrasound device is configured to validate one or more identified structure candidates that match the particular type of structures based on the selective structure detection corresponding to the projection of the three-dimensional (3D) ultrasound volume in the at least one other spatial direction.

4. The system of claim 3 , wherein the ultrasound device is configured to validate each of the one or more identified structure candidates based on spatial matching with a corresponding identified structure candidate determined based on the selective structure detection corresponding to the projection of the three-dimensional (3D) ultrasound volume in the at least one other spatial direction.

5. The system of claim 1 , wherein the ultrasound device is configured to:

generate based on the volumetric ultrasound dataset, at least one other three-dimensional (3D) ultrasound volume; and

apply the selective structure detection to the at least one other three-dimensional (3D) ultrasound volume.

6. The system of claim 5 , wherein the ultrasound device is configured to:

determine an overlapping region between the three-dimensional (3D) ultrasound volume and the at least one other three-dimensional (3D) ultrasound volume; and

validate one or more identified structure candidates that match the particular type of structures, that are located with the overlapping region, based on the selective structure detection applied to the at least one other three-dimensional (3D) ultrasound volume.

7. The system of claim 6 , wherein the ultrasound device is configured to validate each of the one or more identified structure candidates based on spatial matching with a corresponding identified structure candidate, that is within the overlapping region, determined based on the selective structure detection corresponding to the projection of the three-dimensional (3D) ultrasound volume in the at least one other spatial direction.

8. The system of claim 1 , wherein the ultrasound device is configured to adjust the two-dimensional (2D) structure detection based on an adaptive learning algorithm.

9. The system of claim 1 , wherein the ultrasound device is configured to adjust the three-dimensional (3D) structure detection based on an adaptive learning algorithm.

10. A method, comprising:

generating volumetric ultrasound dataset, based on echo ultrasound signals;

generating based on the volumetric ultrasound dataset, a three-dimensional (3D) ultrasound volume; and

applying selective structure detection to the three-dimensional (3D) ultrasound volume, the selective structure detection comprising:

generating based on a projection of the three-dimensional (3D) ultrasound volume in a particular spatial direction, a two-dimensional (2D) image;

applying two-dimensional (2D) structure detection to the two-dimensional (2D) image, wherein the two-dimensional (2D) structure detection is configured to identify structure candidates associated with a particular type of structures, the particular type of structures comprising abnormal tissue structures in a particular body part;

selecting for each identified structure candidate, a corresponding local volume within the three-dimensional (3D) ultrasound volume;

applying three-dimensional (3D) structure detection to each selected local volume, wherein:

the three-dimensional (3D) structure detection is configured to compare characteristics associated with the structure candidate with the particular type of structures, and

the three-dimensional (3D) structure detection is applied at full-resolution to each selected local volume; and

identifying based on applying the three-dimensional (3D) structure detection, one or more structure candidates that match the particular type of structures.

11. The method of claim 10 , comprising applying the selective structure detection based on a projection of the three-dimensional (3D) ultrasound volume in at least one other spatial direction.

12. The method of claim 11 , comprising validating one or more identified structure candidates that match the particular type of structures based on the selective structure detection corresponding to the projection of the three-dimensional (3D) ultrasound volume in the at least one other spatial direction.

13. The method of claim 12 , comprising validating each of the one or more identified structure candidates based on spatial matching with a corresponding identified structure candidate determined based on the selective structure detection corresponding to the projection of the three-dimensional (3D) ultrasound volume in the at least one other spatial direction.

14. The method of claim 10 , comprising:

generating based on the volumetric ultrasound dataset, at least one other three-dimensional (3D) ultrasound volume; and

applying the selective structure detection to the at least one other three-dimensional (3D) ultrasound volume.

15. The method of claim 14 , comprising:

determine an overlapping region between the three-dimensional (3D) ultrasound volume and the at least one other three-dimensional (3D) ultrasound volume; and

validate one or more identified structure candidates that match the particular type of structures, that are located with the overlapping region, based on the selective structure detection applied to the at least one other three-dimensional (3D) ultrasound volume.

16. The method of claim 15 , comprising validating each of the one or more identified structure candidates based on spatial matching with a corresponding identified structure candidate, that is within the overlapping region, determined based on the selective structure detection corresponding to the projection of the three-dimensional (3D) ultrasound volume in the at least one other spatial direction.

17. The method of claim 10 , comprising adjusting the two-dimensional (2D) structure detection based on an adaptive learning algorithm.

18. The method of claim 10 , comprising adjusting the three-dimensional (3D) structure detection based on an adaptive learning algorithm.

19. The method of claim 10 , wherein the two-dimensional (2D) structure detection comprises convolutional neural network (CNN) based detection.

20. The method of claim 10 , wherein the three-dimensional (3D) structure detection comprises convolutional neural network (CNN) based detection.