Resolving and steering decision foci in machine learning-based vascular imaging

1. A system for determining a relative importance of image features of a vascular medical image, the system comprising:

a medical image database comprising a plurality of vascular medical images; and

at least one processor configured to:

analyze the plurality of vascular medical images and automatically generate at least one diagnostic rule corresponding to a common diagnosis of a subset of the plurality of vascular medical images based on a plurality of image features common to the subset of vascular medical images,

receive a current vascular medical image comprising the plurality of image features,

compute an overall diagnostic metric for the current vascular medical image by applying the at least one automatically-generated diagnostic rule to the current vascular medical image,

identify, in the current vascular medical image, a relative importance of each of the plurality of image features to the computed overall diagnostic metric by backpropagation of the computed overall diagnostic metric into contribution of each of the plurality of image features to the computation of the computed overall diagnostic metric, and

provide the current vascular medical image including an indication of the relative importance of each of the plurality of image features in the current vascular medical image.

2. The system according to claim 1 , wherein the at least one processor is further configured to generate the at least one diagnostic rule by executing either a machine learning algorithm, a deep learning algorithm, or an automatic intelligence algorithm.

3. The system according to claim 1 , wherein the overall diagnostic metric is a risk score corresponding to a risk of an acute coronary event, a risk of peripheral vascular disease, a risk of coronary plaque, a risk of coronary vascular disease, or a stenosis level characterization.

4. The system according to claim 1 , wherein the at least one processor is further configured to identify the relative importance of each of the plurality of image features to the computed overall diagnostic metric in a form of a heatmap.

5. The system according to claim 1 , wherein the at least one processor is further configured to:

receive user input from a user input device indicative of at least one of the plurality of image features,

either i) re-compute the overall diagnostic metric for the current vascular medical image by changing the relative importance of the least one of the plurality of image features and re-applying the at least one diagnostic rule to the current vascular medical image or to ii) change the relative importance of the at least one of the plurality of image features,

re-analyze the plurality of vascular medical images to automatically generate a revised diagnostic rule corresponding to a common diagnosis of a selection of the vascular medical images based on a plurality of image features common to the selection of the vascular medical images,

compute a revised overall diagnostic metric for the current vascular medical image by applying the revised automatically-generated diagnostic rule to the current vascular medical image, and

identify, in the current vascular medical image, the relative importance of each of the plurality of image features to the computed revised overall diagnostic metric.

6. The system according to claim 5 , wherein to change the relative importance of the at least one of the plurality of image features, the at least one processor is further configured to increase, decrease, or neglect the relative importance of the at least one of the plurality of image features.

7. The system according to claim 5 , wherein the at least one processor is further configured to automatically generate the at least one diagnostic rule by weighting each of the plurality of image features and changing the relative importance of the at least one of the plurality of image features by changing a weight of the at least one of the plurality of image features.

8. The system according to claim 5 , wherein the user input device is configured to define a region of the current vascular medical image.

9. The system according to claim 5 , wherein the at least one of the plurality of image features corresponds to one or more image artefacts.

10. The system according to claim 1 , wherein the at least one processor is further configured to identify, in the current vascular medical image, the relative importance of each of the plurality of image features to the computed overall diagnostic metric based on either i) a sensitivity of the computed overall diagnostic metric to a variation of each of the plurality of image features or ii) a change in entropy of the computed overall diagnostic metric resulting from a variation of each of the plurality of image features.

11. The system according to claim 1 , wherein the plurality of image features corresponds to one or more of plaque, calcium, a presence of ruptured plaque, a thrombus in a lesion, a presence of fat in a plaque region adjacent to a lumen, and/or includes one or more of a computed lumen area, a computed plaque area, a computed lesion dimension, a computed distribution, or proportion of calcium in a lesion.

12. The system according to claim 1 , wherein the plurality of vascular medical images and the current vascular medical image are either i) IVUS, ii) OCT, iii) (C)CTA, or iv) angiography images.

13. The system according to claim 1 , wherein the at least one processor is further configured to:

compute a measure of uncertainty of the overall diagnostic metric for the current vascular medical image, and

identify said uncertainty in the current vascular medical image.

14. The system according to claim 1 , wherein the backpropagation propagates the overall diagnostic metric back into a relevance score for each of the image features, the relevance score indicating relevance of the corresponding image feature to the computation of the overall diagnostic metric.

15. The system according to claim 1 , wherein the at least one processor is further configured to:

based on the backpropagation of the computed overall diagnostic metric, detect an error in the identified relative importance of an image feature of the plurality of image features to the overall diagnostic metric;

modify the relative importance of the image feature based on the error; and

re-compute the overall diagnostic metric based on the modified relative importance of the image feature.

16. A method of determining a relative importance of image features of a vascular medical image, the method comprising:

analyzing a plurality of vascular medical images and automatically generating at least one diagnostic rule corresponding to a common diagnosis of a subset of the vascular medical images based on a plurality of image features common to the subset of the vascular medical images;

providing a current vascular medical image comprising the plurality of image features;

computing an overall diagnostic metric for the current vascular medical image by applying the at least one automatically-generated diagnostic rule to the current vascular medical image;

identifying, in the current vascular medical image, the relative importance of each of the plurality of image features to the computed overall diagnostic metric by backpropagating the computed overall diagnostic metric into contribution of each of the plurality of image features to the computation of the computed overall diagnostic metric; and

providing the current vascular medical image including an indication of the relative importance of each of the plurality of image features in the current vascular medical image.

17. The method according to claim 16 , further comprising:

based on the backpropagation of the computed overall diagnostic metric, detecting an error in the identified relative importance of an image feature of the plurality of image features to the overall diagnostic metric;

modifying the relative importance of the image feature based on the error; and

re-computing the overall diagnostic metric based on the modified relative importance of the image feature.

18. A non-transitory computer-readable storage medium having stored a computer program comprising instructions, which, when executed by a processor, cause the processor to:

analyze a plurality of vascular medical images to automatically generate at least one diagnostic rule corresponding to a common diagnosis of a subset of the vascular medical images based on a plurality of image features common to the subset of the vascular medical images;

receive a current vascular medical image comprising the plurality of image features;

compute an overall diagnostic metric for the current vascular medical image by applying the at least one automatically-generated diagnostic rule to the current vascular medical image;

identify, in the current vascular medical image, the relative importance of each of the plurality of image features to the computed overall diagnostic metric by backpropagation of the computed overall diagnostic metric into contribution of each of the plurality of image features to the computation of the computed overall diagnostic metric; and

provide the current vascular medical image including an indication of the relative importance of each of the plurality of image features in the current vascular medical image.

19. The A non-transitory computer-readable storage medium according to claim 18 , wherein the backpropagation propagates the overall diagnostic metric back into a relevance score for each of the image features, the relevance score indicating relevance of the corresponding image feature to the computation of the overall diagnostic metric.

20. The A non-transitory computer-readable storage medium according to claim 18 , wherein, the instructions, when executed by a processor, further cause the processor to:

based on the backpropagation of the computed overall diagnostic metric, detect an error in the identified relative importance of an image feature of the plurality of image features to the overall diagnostic metric;

modify the relative importance of the image feature based on the error; and

re-compute the overall diagnostic metric based on the modified relative importance of the image feature.