System and method for analyzing three-dimensional image data

1. A method for determining a malignancy likelihood score for breast tissue of a patient, the method comprising:

receiving a plurality of two-dimensional images of the breast tissue, the two-dimensional images being derived from a three-dimensional image of the breast tissue;

providing each of the two-dimensional images to a first model comprising a first trained neural network that is trained to identify regions of interest; and

receiving a first identified region of interest from the first model corresponding to a first image of the plurality of two-dimensional images;

receiving a second identified region of interest from the first model corresponding to a second image of the plurality of two-dimensional images;

generating a synthetic two-dimensional image by populating a pixel array of the synthetic two-dimensional image based on at least the first identified region of interest and the second identified region of interest;

providing the synthetic two-dimensional image to a second model comprising a second trained neural network;

receiving a malignancy likelihood score from the second model; and

outputting a report including the malignancy likelihood score to at least one of a memory or a display.

2. The method of claim 1 , comprising:

determining that the first region of interest and the second region of interest overlap;

determining that the first region of interest has a higher relevancy score than the second region of interest; and

in response to determining the first region of interest has a higher relevancy score, including at least a portion of the first region of interest in the synthetic two-dimensional image.

3. The method of claim 2 , wherein the first region of interest is associated with a first two-dimensional image included in the plurality of two-dimensional images and the second region of interest is associated with a second two-dimensional image included in the plurality of two-dimensional images.

4. The method of claim 1 , wherein each of the first and second regions of interest comprises an array of pixels, each pixel comprising an intensity value, and a relevancy score.

5. The method of claim 4 , wherein the generating comprises:

setting a first intensity value included in a first pixel included in the synthetic two-dimensional image equal to a second intensity value included in the first region of interest.

6. The method of claim 4 , wherein the generating comprises:

setting a first intensity value included in a first pixel included in the synthetic two-dimensional image equal to a second intensity value included in a third two-dimensional image included in the plurality of two-dimensional images, the second intensity value being included in a second pixel not associated with any region of interest identified by the first model.

7. The method of claim 6 , wherein the second third two-dimensional image is associated with at least one region of interest identified by the first model.

8. The method of claim 4 , wherein the generating comprises:

determining a first coverage area associated with the first region of interest and comprising a second array of pixels;

determining a second coverage area associated with the second region of interest and comprising a third array of pixels; and

determining a first intensity value of a first pixel included in the synthetic two-dimensional image based on a second intensity value included in the first coverage area and a third intensity value included in the second coverage area.

9. The method of claim 8 , wherein the first coverage area includes at least one pixel not included in the first region of interest and the second coverage area includes at least one pixel not included in the second region of interest.

10. The method of claim 8 , wherein the determining the first intensity value of the first pixel comprises:

setting the first intensity value equal to the sum of the second intensity value multiplied by a first weight plus the third intensity value multiplied by a second weight.

11. The method of claim 1 , wherein the generating comprises:

determining that the first region of interest and the second region of interest overlap;

determining that the first region of interest has a higher relevancy score than the second region of interest; and

in response to determining the first region of interest has a higher relevancy score, not including any portion of the second region of interest in the synthetic two-dimensional image.

12. The method of claim 1 , wherein the three-dimensional image is generated using digital breast tomosynthesis.

13. The method of claim 1 , wherein the first trained neural network comprises a first subnetwork, and the second trained neural network comprises a second subnetwork, the first subnetwork and the second subnetwork including the same number of layers and filters.

14. The method of claim 13 , wherein the second trained neural network was trained using a set of weight values of the first neural network as initial weight values.

15. The method of claim 1 further comprising:

removing at least one region of interest from a number of regions of interest received from the first model; and

generating a second synthetic two-dimensional image based on the number of regions of interest and at least one of the plurality of two-dimensional images.

16. The method of claim 1 , wherein the first model is trained based on an image dataset comprising two-dimensional full-field digital mammography images annotated by a medical practitioner.

17. The method of claim 16 , wherein each two-dimensional image included in the plurality of two-dimensional images of the breast tissue is a slice included in a three-dimensional digital breast tomosynthesis volume.

18. A system for determining a malignancy likelihood score for breast tissue of a patient, the system comprising:

a memory configured to store a plurality of two-dimensional images of the breast tissue, the two-dimensional images being derived from a three-dimensional image of the breast tissue;

a processor configured to access the memory and to:

provide each two-dimensional image to a first model comprising a first trained neural network that is trained to identify regions of interest;

receive a first identified region of interest from the first model corresponding to a first image of the plurality of two-dimensional images;

receive a second identified region of interest from the first model corresponding to a second image of the plurality of two-dimensional images;

generate a synthetic two-dimensional image by populating a pixel array of the synthetic two-dimensional image based on at least the first identified region of interest and the second identified region of interest;

provide the synthetic two-dimensional image to a second model comprising a second trained neural network;

determine a malignancy likelihood score using the second model; and

a display configured to display a report including the malignancy likelihood score.

19. The system of claim 18 , wherein, to generate the synthetic two-dimensional image, the processor is further to:

determine that the first region of interest and the second region of interest overlap;

determine that the first region of interest has a higher relevancy score than the second region of interest; and

in response to determining the first region of interest has a higher relevancy score, include at least a portion of the first region of interest in the synthetic two-dimensional image.

20. The system of claim 19 , wherein the first region of interest is associated with a first two-dimensional image included in the plurality of two-dimensional images and the second region of interest is associated with a second two-dimensional image included in the plurality of two-dimensional images.

21. The system of claim 18 , wherein each of the first and second regions of interest comprises an array of pixels, each pixel comprising an intensity value, and a relevancy score.

22. The system of claim 21 , wherein, to generate the synthetic two-dimensional image, the processor is further to:

set a first intensity value included in a first pixel included in the synthetic two-dimensional image equal to a second intensity value included in the first region of interest.

23. The system of claim 21 , wherein, to generate the synthetic two-dimensional image, the processor is further to:

set a first intensity value included in a first pixel included in the synthetic two-dimensional image equal to a second intensity value included in a third two-dimensional image included in the plurality of two-dimensional images, the second intensity value being included in a second pixel not associated with any region of interest identified by the first model.

24. The system of claim 21 , wherein, to generate the synthetic two-dimensional image, the processor is further to:

determine a first coverage area associated with the first region of interest comprising a second array of pixels;

determine a second coverage area associated with the second region of interest and comprising a third array of pixels; and

determine a first intensity value of a first pixel included in the synthetic two-dimensional image based on a second intensity value included in the first coverage area and a third intensity value included in the second coverage area.

25. The system of claim 24 , wherein the first coverage area includes at least one pixel not included in the first region of interest and the second coverage area includes at least one pixel not included in the second region of interest.

26. The system of claim 24 , wherein, to determine the first intensity value of the first pixel, the processor is further to:

set the first intensity value equal to the sum of the second intensity value multiplied by a first weight plus the third intensity value multiplied by a second weight.

27. The system of claim 18 , wherein, to generate the synthetic two-dimensional image, the processor is further to:

determine that the first region of interest and the second region of interest overlap;

determine that the first region of interest has a higher relevancy score than the second region of interest; and

in response to determining the first region of interest has a higher relevancy score, not include any portion of the second region of interest in the synthetic two-dimensional image.

28. The system of claim 18 , wherein the three-dimensional image is generated using digital breast tomosynthesis and wherein each two-dimensional image included in the plurality of two-dimensional images of the breast tissue is a slice included in a three-dimensional digital breast tomosynthesis volume.

29. The system of claim 18 , wherein the first trained neural network comprises a first subnetwork, and the second trained neural network comprises a second subnetwork, the first subnetwork and the second subnetwork including the same number of layers and filters, and wherein the second trained neural network was trained using a set of weight values of the first neural network as initial weight values.

30. A method for determining a malignancy likelihood score for breast tissue of a patient, the method comprising:

receiving a plurality of two-dimensional images of the breast tissue, the two-dimensional images being derived from a three-dimensional image of the breast tissue;

for each two-dimensional image:

providing the two-dimensional image to a first model comprising a first trained neural network; and

receiving a number of indicators from the first model, each indicator being associated with a two-dimensional image included in the plurality of two-dimensional images, wherein each indicator comprises an array of pixels, each pixel comprising an intensity value, and a relevancy score;

generating a synthetic two-dimensional image based on the number of indicators and at least one of the plurality of two-dimensional images, wherein the generating comprises:

determining a first coverage area associated with a first indicator included in the number of indicators and comprising a second array of pixels,

determining a second coverage area associated with a second indicator included in the number of indicators and comprising a third array of pixels, and

determining a first intensity value of a first pixel included in the synthetic two-dimensional image based on a second intensity value included in the first coverage area and a third intensity value included in the second coverage area;

providing the synthetic two-dimensional image to a second model comprising a second trained neural network;

receiving a malignancy likelihood score from the second model; and

outputting a report including the malignancy likelihood score to at least one of a memory or a display.

31. The method of claim 30 , wherein the first coverage area includes at least one pixel not included in the first indicator and the second coverage area includes at least one pixel not included in the second indicator.

32. The method of claim 30 , wherein the determining the first intensity value of the first pixel comprises:

setting the first intensity value equal to the sum of the second intensity value multiplied by a first weight plus the third intensity value multiplied by a second weight.