Spatial distribution of pathological image patterns in 3D image data
A method and for quantifying a three-dimensional medical image volume are provided. An embodiment of the method includes: providing a two-dimensional representation image based on the medical image volume; defining a region of interest in the two-dimensional representation image; generating a feature signature for the region of interest; defining a plurality of two-dimensional image patches in the medical image volume; calculating, for each of the image patches, a degree of similarity between the region of interest and the respective image patch on the basis of the feature signature; visualizing the degrees of similarities.
1. A method for quantifying a three-dimensional medical image volume comprising:
providing a two-dimensional representation image based on the three-dimensional medical image volume;
defining a region of interest in the two-dimensional representation image;
generating a feature signature for the region of interest;
defining a plurality of two-dimensional image patches in the three-dimensional medical image volume;
calculating, for each of the plurality of two-dimensional image patches, a degree of similarity between the region of interest and the respective image patch based on the feature signature; and
visualizing the degrees of similarities, wherein
the three-dimensional medical image volume contains a stack of two-dimensional sequential images,
the step of providing the two-dimensional representation image comprises selecting one of the two-dimensional sequential images as the two-dimensional representation image, and
the step of defining of the plurality of two-dimensional image patches comprises defining a plurality of image patches in each of the two-dimensional sequential images.
2. The method of claim 1 , wherein
in the step of defining the plurality of image patches, a size of the image patches is defined based on a size of the region of interest.
3. The method of claim 1 , wherein the step of generating the feature signature for the region of interest is performed using a trained machine-learning algorithm.
4. The method of claim 1 , wherein
the step of visualizing the degrees of similarities comprises generating a two-dimensional rendering based on the degrees of similarity.
5. The method of claim 1 , wherein
the defining of the region of interest is carried out manually by a user.
6. The method of claim 1 , wherein
the step of providing the two-dimensional representation image comprises selecting the two-dimensional representation image from the three-dimensional medical image volume manually by a user.
7. The method of claim 1 , wherein
the degrees of similarity span the three-dimensional medical image volume in three dimensions.
8. The method of claim 1 , further with the steps of:
generating a segmentation mask for the three-dimensional medical image volume; and
applying the segmentation mask in the step of defining the plurality of two-dimensional image patches so that image patches are only defined inside the segmentation mask.
9. The method of claim 1 , further with the steps of:
generating a segmentation mask for the three-dimensional medical image volume; and
applying the segmentation mask in the step of calculating the degrees of similarity so that degrees of similarities are only calculated for image patches inside of the segmentation mask.
10. The method of claim 1 , further with the steps of:
generating a segmentation mask for the three-dimensional medical image volume; and
applying the segmentation mask in the step of visualizing the degrees of similarity so that only degrees of similarity for image patches inside of the segmentation mask are being visualized.
11. The method of claim 1 , wherein, in the step of visualizing the degrees of similarity, the degrees of similarity are visualized using a color-code.
12. The method of claim 1 , wherein
the defining of the region of interest is based on identifying an anatomical feature in the two-dimensional representation image, and wherein the anatomical feature is indicative of a pathological condition of a patient.
13. The method of claim 1 , further comprising the step of
comparing the degrees of similarity to a threshold,
wherein, in the step of visualizing the degrees of similarity, only degrees of similarity above the threshold are visualized.
14. A method for quantifying a three-dimensional medical image volume comprising:
providing a two-dimensional representation image based on the three-dimensional medical image volume;
defining a region of interest in the two-dimensional representation image;
generating a feature signature for the region of interest;
defining a plurality of two-dimensional image patches in the three-dimensional medical image volume;
calculating, for each of the plurality of two-dimensional image patches, a degree of similarity between the region of interest and the respective image patch based on the feature signature; and
visualizing the degrees of similarities, wherein
the step of calculating the degrees of similarity comprises, for each image patch, generating a feature signature for the respective image patch and comparing it to the feature signature of the region of interest.
15. The method of claim 14 , wherein
the step of generating the feature signatures for each of the plurality of two-dimensional image patches is performed using a trained machine-learning algorithm.
16. A method for quantifying a three-dimensional medical image volume comprising:
providing a two-dimensional representation image based on the three-dimensional medical image volume;
defining a region of interest in the two-dimensional representation image;
generating a feature signature for the region of interest;
defining a plurality of two-dimensional image patches in the three-dimensional medical image volume;
calculating, for each of the plurality of two-dimensional image patches, a degree of similarity between the region of interest and the respective image patch based on the feature signature; and
visualizing the degrees of similarities, wherein
the step of visualizing the degrees of similarities comprises:
generating a semi-transparent two-dimensional rendering based on the degrees of similarity, and
overlaying the semi-transparent two-dimensional rendering over a corresponding two-dimensional rendering of the three-dimensional medical image volume.
17. A system for quantifying a three-dimensional medical image volume, comprising:
an interface configured to,
provide a two-dimensional representation image based on the three-dimensional medical image volume; and
at least one processor configured to cause the system to,
define a region of interest in the two-dimensional representation image, generate a feature signature for the region of interest,
define a plurality of two-dimensional image patches in the three-dimensional medical image volume,
calculate, for each of the plurality of two-dimensional image patches, a degree of similarity between the region of interest and the respective image patches based on the feature signature, and
visualize the degrees of similarities, wherein
the three-dimensional medical image volume contains a stack of two-dimensional sequential images,
the at least one processor is further configured to cause the system to provide the two-dimensional representation image by selecting one of the two-dimensional sequential images as the two-dimensional representation image, and
the at least one processor is further configured to cause the system to define the plurality of two-dimensional image patches by defining a plurality of image patches in each of the two-dimensional sequential images.
18. A system for quantifying a three-dimensional medical image volume, comprising:
an interface configured to,
provide a two-dimensional representation image based on the three-dimensional medical image volume; and
at least one processor configured to cause the system to,
define a region of interest in the two-dimensional representation image, generate a feature signature for the region of interest,
define a plurality of two-dimensional image patches in the three-dimensional medical image volume,
calculate, for each of the plurality of two-dimensional image patches, a degree of similarity between the region of interest and the respective image patches based on the feature signature, and
visualize the degrees of similarities, wherein
the at least one processor is further configured to cause the system to calculate the degrees of similarity by, for each image patch, generating a feature signature for the respective image patch and comparing it to the feature signature of the region of interest.
19. A system for quantifying a three-dimensional medical image volume, comprising:
an interface configured to,
provide a two-dimensional representation image based on the three-dimensional medical image volume; and
at least one processor configured to cause the system to,
define a region of interest in the two-dimensional representation image, generate a feature signature for the region of interest,
define a plurality of two-dimensional image patches in the three-dimensional medical image volume,
calculate, for each of the plurality of two-dimensional image patches, a degree of similarity between the region of interest and the respective image patches based on the feature signature, and
visualize the degrees of similarities, wherein
the at least one processor is further configured cause the system to visualize the degrees of similarities by
generating a semi-transparent two-dimensional rendering based on the degrees of similarity, and
overlaying the semi-transparent two-dimensional rendering over a corresponding two-dimensional rendering of the three-dimensional medical image volume.