Synthesis of contrast enhanced medical images
Systems and methods for generating a synthesized contrast enhanced medical image are provided. An input medical image is received. A synthesized contrast enhanced medical image is generated based on the input medical image using a trained machine learning based generator network. The synthesized contrast enhanced medical image includes one or more synthesized contrast enhanced regions of pathological tissue. The synthesized contrast enhanced medical image is output.
1. A computer implemented method comprising:
receiving an input medical image of healthy tissue;
generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network, the synthesized contrast enhanced medical image comprising one or more synthesized contrast enhanced regions of pathological tissue; and
outputting the synthesized contrast enhanced medical image.
2. The computer implemented method of claim 1 , wherein generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network comprises:
generating the synthesized contrast enhanced medical image comprising a random pattern of the one or more synthesized contrast enhanced regions.
3. The computer implemented method of claim 1 , further comprising receiving an enhancement map defining one or more regions of the input medical image to be contrast enhanced, wherein generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network comprises:
generating the synthesized contrast enhanced medical image based on the enhancement map, wherein the one or more synthesized contrast enhanced regions of pathological tissue in the synthesized contrast enhanced medical image are generated in the one or more regions defined by the enhancement map.
4. The computer implemented method of claim 3 , further comprising:
generating the enhancement map based on a thresholded strain map.
5. The computer implemented method of claim 3 , further comprising:
generating the enhancement map based on a segmentation mask associated with an anatomically similar patient.
6. The computer implemented method of claim 1 , further comprising receiving a segmentation mask of an anatomical structure depicted in the input medical image, wherein generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network comprises:
generating the synthesized contrast enhanced medical image based on the segmentation mask using the trained machine learning based generator network.
7. The computer implemented method of claim 1 , further comprising:
determining a segmentation mask of an anatomical structure depicted in the synthesized contrast enhanced medical image as a segmentation mask of the anatomical structure in the input medical image.
8. The computer implemented method of claim 1 , further comprising:
generating a segmentation mask of an anatomical structure depicted in the synthesized contrast enhanced medical image by:
generating a paired non-enhanced medical image of the anatomical structure depicted in the synthesized contrast enhanced medical image using another generator network; and
determining a segmentation mask of the generated paired non-enhanced medical image.
9. The computer implemented method of claim 1 , further comprising:
training a machine learning based system for performing a medical image analysis task based on the synthesized contrast enhanced medical image.
10. The computer implemented method of claim 1 , wherein the input medical image is a cine MRI (magnetic resonance imaging) image and the synthesized contrast enhanced medical image is a synthesized LGE (late gadolinium enhancement) MRI image.
11. An apparatus comprising:
means for receiving an input medical image of healthy tissue;
means for generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network, the synthesized contrast enhanced medical image comprising one or more synthesized contrast enhanced regions of pathological tissue; and
means for outputting the synthesized contrast enhanced medical image.
12. The apparatus of claim 11 , wherein the means for generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network comprises:
means for generating the synthesized contrast enhanced medical image comprising a random pattern of the one or more synthesized contrast enhanced regions.
13. The apparatus of claim 11 , further comprising means for receiving an enhancement map defining one or more regions of the input medical image to be contrast enhanced, wherein the means for generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network comprises:
means for generating the synthesized contrast enhanced medical image based on the enhancement map, wherein the one or more synthesized contrast enhanced regions of pathological tissue in the synthesized contrast enhanced medical image are generated in the one or more regions defined by the enhancement map.
14. The apparatus of claim 13 , further comprising:
means for generating the enhancement map based on a thresholded strain map or based on a segmentation mask associated with an anatomically similar patient.
15. A non-transitory computer readable medium storing computer program instructions, the computer program instructions when executed by a processor cause the processor to perform operations comprising:
receiving an input medical image of healthy tissue;
generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network, the synthesized contrast enhanced medical image comprising one or more synthesized contrast enhanced regions of pathological tissue; and
outputting the synthesized contrast enhanced medical image.
16. The non-transitory computer readable medium of claim 15 , the operations further comprising receiving an enhancement map defining one or more regions of the input medical image to be contrast enhanced, wherein generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network comprises:
generating the synthesized contrast enhanced medical image comprising the one or more synthesized contrast enhanced regions of pathological tissue in the one or more regions defined by the enhancement map.
17. The non-transitory computer readable medium of claim 15 , the operations further comprising receiving a segmentation mask of an anatomical structure depicted in the input medical image, wherein generating a synthesized contrast enhanced medical image based on the input medical image of the healthy tissue using a trained machine learning based generator network comprises:
generating the synthesized contrast enhanced medical image based on the segmentation mask using the trained machine learning based generator network.
18. The non-transitory computer readable medium of claim 15 , the operations further comprising:
determining a segmentation mask of an anatomical structure depicted in the synthesized contrast enhanced medical image as a segmentation mask of the anatomical structure in the input medical image.
19. The non-transitory computer readable medium of claim 15 , the operations further comprising:
generating a segmentation mask of an anatomical structure depicted in the synthesized contrast enhanced medical image by:
generating a paired non-enhanced medical image of the anatomical structure depicted in the synthesized contrast enhanced medical image using another generator network; and
determining a segmentation mask of the generated paired non-enhanced medical image.
20. The non-transitory computer readable medium of claim 15 , the operations further comprising:
training a machine learning based system for performing a medical image analysis task based on the synthesized contrast enhanced medical image.