EP-4571645-B1 - DETERMINING A LOCATION AT WHICH A GIVEN FEATURE IS REPRESENTED IN MEDICAL IMAGING DATA

Inventors

• YEREBAKAN, HALID
• SHINAGAWA, YOSHIHISA
• IYER, Kritika
• HERMOSILLO VALADEZ, GERARDO

Dates

Publication Date

20260506

Application Date

20231211

Claims (15)

1. A computer implemented method of determining a location at which a given feature is represented in target medical imaging data (330), the medical imaging data comprising an array of elements having respective values and representing respective locations, the method comprising: obtaining (112) an initial descriptor for an initial location (710) in the target medical imaging data (330), the initial descriptor (710) being representative of values of elements of the target medical imaging data (330) located relative to the initial location according to a first predefined pattern; determining (114), based on an input of data representative of the initial descriptor to a trained machine learning model (715), an approximate location in the target medical imaging data (330) at which the given feature is represented; determining (122), based on the approximate location, a plurality of candidate locations (442) in the target medical imaging data (330); obtaining (124) candidate descriptors for each of the plurality of candidate locations (442), each candidate descriptor being representative of values of elements of the target medical imaging data located relative to the candidate location according to a second predefined pattern (440); obtaining (126) a reference descriptor for a reference location in reference medical imaging data (220), the reference medical imaging data (220) comprising one or more sets of reference medical imaging data, the reference location being, for each of the one or more sets of reference medical imaging data, a location in the set of reference medical imaging data at which the given feature is represented, the reference descriptor being representative of, for each of the one or more sets of reference medical imaging data, values of elements of the set of reference medical imaging data located relative to the reference location according to the second predefined pattern (440); for each of the plurality of candidate locations (442), comparing (128) the reference descriptor and the candidate descriptor for the candidate location to obtain a similarity metric; selecting (130) a candidate location from among the plurality of candidate locations based on the calculated similarity metrics; and determining (123) the location at which the given feature (226) is represented in the target medical imaging data (330) based on the selected candidate location.
2. The method of claim 1, comprising: determining, based on the determined location at which the given feature is represented in the target medical imaging data, a plurality of further candidate locations in the target medical imaging data, wherein a distance between the further candidate locations is less than a distance between the candidate locations; obtaining further candidate descriptors for each of the plurality of further candidate locations, each further candidate descriptor being representative of values of elements of the target medical imaging data located relative to the further candidate location according to the second predefined pattern; for each of the plurality of further candidate locations, comparing the reference descriptor and the further candidate descriptor for the candidate location to obtain a further similarity metric; selecting a further candidate location from among the plurality of further candidate locations based on the further similarity metrics; and determining a refined location at which the given feature is represented in the target medical imaging data based on the selected further candidate location.
3. The method of claim 1 or claim 2, comprising performing an image registration process using the location at which the given feature is represented in the target medical imaging data.
4. The method of claim 3, further comprising using the computer implemented method of claim 1 to determine a location at which a further feature is represented in the target medical imaging data, wherein performing the registration process comprises performing the registration process using the location at which the further feature is represented in the target medical imaging data.
5. The method of any one of claim 1 to claim 4, wherein determining the location at which the given feature is represented comprises determining, as the location at which the given feature is represented in the target medical imaging data, the selected candidate location.
6. The method of any one of claim 1 to claim 5, wherein the one or more sets of reference medical imaging data comprises a plurality of sets of reference medical imaging data, and the reference descriptor is an averaged reference descriptor obtained from the plurality of sets of reference medical imaging data.
7. The method of any one of claim 1 to claim 6, wherein determining the approximate location comprises: generating, based on an input of data representative of the initial descriptor to the trained machine learning model, an initial set of coordinates representing an initial body location in a template body, the initial body location in the template body corresponding to a body location, in a body at least a portion of which is represented by the target medical imaging data, represented at the initial location in the target medical imaging data; and based on the initial set of coordinates, a set of feature coordinates representing a location of the given feature in the template body, and the initial location, determining the approximate location.
8. The method of claim 7, wherein determining the approximate location comprises calculating a vector between the initial set of coordinates and the set of feature coordinates, and calculating the approximate location based on the initial location and the vector.
9. The method of claim 8, wherein calculating the approximate location comprises adding the vector to a vector representation of the initial location.
10. The method of any one of claim 7 to claim 9, wherein the method comprises a refining process to refine the approximate location, the refining process comprising: determining, based on the initial set of coordinates and the set of feature coordinates, a direction; determining, based on the initial location and the direction, a further initial location in the target medical imaging data; obtaining a further descriptor for the further initial location, the further descriptor being representative of values of elements of the target medical imaging data located relative to the further initial location according to the first predefined pattern; generating, based on the input of the data representative of the further descriptor to the trained machine learning model, a further initial set of coordinates representing a further initial body location in the template body, the further initial body location in the template body corresponding to a body location, in the body at least a portion of which is represented by the given medical imaging data, represented at the further initial location in the target medical imaging data; and based on the further initial set of coordinates, the set of feature coordinates representing the location of the given feature in the template body, and the further initial location, determining the approximate location.
11. The method of claim 10, wherein determining the direction comprises determining, as the direction, a direction of a vector between the initial set of coordinates and the set of feature coordinates.
12. The method of any one of claim 7 to claim 11, wherein the trained machine learning model has been trained by a training method comprising: providing a machine learning model configured to generate, based on an input of data representative of a given descriptor, a set of coordinates representing a body location in the template body, the given descriptor being representative of values of elements of given medical imaging data located relative to a given location in the given medical imaging data according to the first predefined pattern; providing training data comprising a plurality of training descriptors, each training descriptor being representative of values of elements of a given set of medical imaging data located relative to a given location in the given set of medical imaging data according to the first predefined pattern, the training data further comprising, for each training descriptor, a ground truth set of coordinates associated with a respective body location in the template body; and training the machine learning model based on the training data so as to minimize a loss function between sets of coordinates generated by the machine learning model based on the training descriptors and the corresponding ground truth sets of coordinates.
13. The method of claim 12, wherein providing the training data comprises obtaining one of the plurality of training descriptors by: obtaining a template descriptor for a reference location in template medical imaging data, the reference location in the template medical imaging data being a location in the template medical imaging data at which the respective body location is represented, the template descriptor being representative of values of elements of the template medical imaging data located relative to the reference location according to the second predefined pattern; obtaining candidate descriptors for each of a plurality of candidate locations in the given set of medical imaging data, each candidate descriptor being representative of values of elements of the given set of medical imaging data located relative to the candidate location according to the second predefined pattern; for each of the plurality of candidate locations in the given set of medical imaging data, comparing the template descriptor and the candidate descriptor for the candidate location to obtain a template similarity metric; and determining the training descriptor based on the candidate descriptors and the calculated template similarity metrics.
14. Apparatus configured to perform the method of any one of claim 1 to claim 13.
15. A computer program which when executed by a computer causes the computer to perform the method of any one of claim 1 to claim 13.

Description

Technical Field The invention relates to a method and apparatus for determining a location at which a given feature is represented in medical imaging data. Background In medical imaging, it is often important to determine where a certain feature, such as a medical landmark (e.g. the tracheal carina), is represented in a medical image. Landmarks can be used to register different images into one coordinate system. For example, when comparing two images of a patient taken at different times, position and zoom of the images may be different. By registering them into one coordinate system, the images can be compared to evaluate the progress of a disease. Landmarks also have other uses, such as planning radiotherapy, and determining geometrical features (e.g. distances) in a human body. One method for finding landmarks involves comparing a descriptor of the landmark in a reference medical image with candidate descriptors for candidate locations in the target medical image, and identifying the candidate location whose descriptor is the most similar to the reference descriptor. However, this requires testing candidate locations across the entire target medical image. Furthermore, the method may identify a candidate location as representing the landmark even if the landmark is not actually represented in the target medical image. As such, this method can be susceptible to returning 'false positive' indications of the location of a particular landmark in a target medical image, which is undesirable. Earlier approaches are known from e.g. US 2023/005136 A1. It is desirable to provide a reliable and efficient method for determining the location of a feature in medical imaging data. Summary According to a first aspect of the present invention, there is a provided a computer implemented method of determining a location at which a given feature is represented in target medical imaging data, the medical imaging data comprising an array of elements having respective values and representing respective locations, the method comprising: obtaining an initial descriptor for an initial location in the target medical imaging data, the initial descriptor being representative of values of elements of the target medical imaging data located relative to the initial location according to a first predefined pattern; determining, based on an input of data representative of the initial descriptor to a trained machine learning model, an approximate location in the target medical imaging data at which the given feature is represented; determining, based on the approximate location, a plurality of candidate locations in the target medical imaging data; obtaining candidate descriptors for each of the plurality of candidate locations, each candidate descriptor being representative of values of elements of the target medical imaging data located relative to the candidate location according to a second predefined pattern; obtaining a reference descriptor for a reference location in reference medical imaging data, the reference medical imaging data comprising one or more sets of reference medical imaging data, the reference location being, for each of the one or more sets of reference medical imaging data, a location in the set of reference medical imaging data at which the given feature is represented, the reference descriptor being representative of, for each of the one or more sets of reference medical imaging data, values of elements of the set of reference medical imaging data located relative to the reference location according to the second predefined pattern; for each of the plurality of candidate locations, comparing the reference descriptor and the candidate descriptor for the candidate location to obtain a similarity metric; selecting a candidate location from among the plurality of candidate locations based on the calculated similarity metrics; and determining the location at which the given feature is represented in the target medical imaging data based on the selected candidate location. Optionally, the method comprises: determining, based on the determined location at which the given feature is represented in the target medical imaging data, a plurality of further candidate locations in the target medical imaging data, wherein a distance between the further candidate locations is less than a distance between the candidate locations; obtaining further candidate descriptors for each of the plurality of further candidate locations, each further candidate descriptor being representative of values of elements of the target medical imaging data located relative to the further candidate location according to the second predefined pattern; for each of the plurality of further candidate locations, comparing the reference descriptor and the further candidate descriptor for the candidate location to obtain a further similarity metric; selecting a further candidate location from among the plurality of further candidate locations based on the f