Methods and systems for identifying slices in medical image data sets

1 . A computer-implemented method comprising:

receiving a first medical image data set;

extracting at least one image descriptor from image data of the first medical image data set, the at least one image descriptor from the image data of the first medical image data set being a first feature vector;

receiving a second medical image data set, the second medical image data set including a plurality of slices and being different from the first medical image data set;

extracting at least one image descriptor from each slice of the plurality of slices of the second medical image data set, the at least one image descriptor from each slice of the plurality of slices of the second medical image data set being a second feature vector;

comparing, with a trained function, the at least one image descriptor of the first medical image data set and the at least one image descriptor from each slice of the plurality of slices of the second medical image data set to generate a degree of similarity between the first medical image data set and each slice of the plurality of slices of the second medical image data set; and

identifying, from the plurality of slices of the second medical image data set, at least one corresponding slice based on the degrees of similarity between the first medical image data set and each slice of the plurality of slices of the second medical image data set, wherein

the trained function is configured to determine degrees of similarity between two-dimensional medical images, and

the trained function applies a learned metric to determine the degrees of similarity between the two-dimensional medical images, the trained function including a deep metric learning network.

2 . The method according to claim 1 , wherein

the first medical image data set comprises a plurality of slices, and

the identifying includes,

identifying, for each of the plurality of slices of the first medical image data set, one corresponding slice of the second medical image data set to determine a slice correspondence between the first medical image data set and the second medical image data set, the degrees of similarity being respectively based on a similarity between an individual slice of the first medical image data set and an individual slice of the second medical image data set.

3 . The method according to claim 2 , further comprising:

determining an anatomical overlap between an image volume of the first medical image data set and an image volume of the second medical image data set based on the identified slice correspondence;

evaluating a degree of comparability of the second medical image data set with the first medical image data set based on the determined anatomical overlap; and

providing the degree of comparability to a user via a user interface.

4 . The method according to claim 2 , further comprising:

receiving an input from a user indicative of a selection of a slice of the first medical image data set to designate a selected slice;

identifying, from the plurality of slices of the second medical image data set, a slice corresponding to the selected slice based on the identified slice correspondence; and

providing the slice corresponding to the selected slice.

5 . The method according to claim 1 , wherein the first medical image data set comprises a plurality of slices and the extracting the at least one image descriptor of the first medical data set includes

extracting an image descriptor from each of the plurality of slices of the first medical image data set.

6 . The method according to claim 1 , wherein:

the first medical image data set is associated with a first medical imaging modality,

the second medical image data set is associated with a second medical imaging modality, and

the first medical imaging modality is based on a first imaging technology different from a second imaging technology of the second medical imaging modality, the first imaging technology being an x-ray imaging technology and the second imaging technology being a magnetic resonance imaging technology.

7 . The method according to claim 1 , wherein:

the first medical image data set comprises a two-dimensional key image indicative of at least one finding previously reported for a patient; and

the second medical image data set comprises an image study a reference image has been extracted from upon reporting the at least one finding for the patient.

8 . The method according to claim 1 , further comprising:

resampling the second medical data set based on the first medical image data set in order to define the plurality of slices in the second medical image data set.

9 . The method according to claim 1 , wherein the trained function is configured to determine degrees of similarities between the two-dimensional medical images by:

comparing a first candidate image to a reference image and comparing a second candidate image to the reference image, the reference image being extracted from one of the first and second medical image data sets and the first and second candidate images being extracted from the respective other one of the first and second medical image data set; and

determining which one of the first and second candidate images has a higher degree of similarity to the reference image.

10 . A system comprising:

an interface configured to receive a first medical image data set and a second medical image data set, the second medical image data set including a plurality of slices and being different from the first medical image data set; and

a computing unit configured to cause the system to,

extract at least one image descriptor from image data of the first medical image data set, the at least one image descriptor from the image data of the first medical image data set being a first feature vector,

extract at least one image descriptor from each slice of the plurality of slices of the second medical image data set, the at least one image descriptor from each slice of the plurality of slices of the second medical image data set being a second feature vector,

compare, with a trained function, the at least one image descriptor of the first medical image data set and the at least one image descriptor from each slice of the plurality of slices of the second medical image data set to generate a degree of similarity between the first medical image data set and each slice of the plurality of slices of the second medical image data set, and

identify, from the plurality of slices of the second medical image data set, at least one corresponding slice based on the degrees of similarity between the first medical image data set and each slice of the plurality of slices of the second medical image data set, wherein

the trained function is configured to determine degrees of similarity between two-dimensional medical images, and

the trained function applies a learned metric to determine the degrees of similarity between the two-dimensional medical images, the trained function including a deep metric learning network.

11 . The system according to claim 10 , wherein

the first medical image data set comprises a plurality of slices, and

the computing unit is configured to cause the system to,

identify, for each of the plurality of slices of the first medical image data set, one corresponding slice of the second medical image data set to determine a slice correspondence between the first medical image data set and the second medical image data set, the degrees of similarity being respectively based on a similarity between an individual slice of the first medical image data set and an individual slice of the second medical image data set.

12 . The system according to claim 11 , wherein the computing unit is configured to cause the system to,

determine an anatomical overlap between an image volume of the first medical image data set and an image volume of the second medical image data set based on the identified slice correspondence;

evaluate a degree of comparability of the second medical image data set with the first medical image data set based on the determined anatomical overlap; and

provide the degree of comparability to a user via a user interface.

13 . The system according to claim 11 , wherein the computing unit is configured to cause the system to,

receive an input from a user indicative of a selection of a slice of the first medical image data set to designate a selected slice;

identify, from the plurality of slices of the second medical image data set, the slice corresponding to the selected slice based on the identified slice correspondence; and

provide the slice corresponding to the selected slice.

14 . The system according to claim 10 , wherein the first medical image data set comprises a plurality of slices and the extracting the at least one image descriptor of the first medical data set includes

extracting an image descriptor from each of the plurality of slices of the first medical image data set.

15 . A non-transitory computer-readable medium comprising program elements, when executed by a computing unit, are configured to cause the method of claim 1 to be performed.