Preprocessing medical image data for machine learning

Patent metadata

Field	Value
Publication number	US-11861839-B2
Application number	US-201917055255-A
Country	US
Kind code	B2
Filing date	May 8, 2019
Priority date	May 14, 2018
Publication date	Jan 2, 2024
Grant date	Jan 2, 2024

Abstract

A system and computer-implemented method are provided for preprocessing medical image data for machine learning. Image data is accessed which comprises an anatomical structure. The anatomical structure in the image data is segmented to obtain a segmentation of the anatomical structure as a delineated part of the image data. A grid is assigned to the delineated part of the image data, the grid representing a partitioning of an exterior and interior of the type of anatomical structure using grid lines, wherein said assigning comprises adapting the grid to fit the segmentation of the anatomical structure in the image data. A machine learning algorithm is then provided with an addressing to the image data in the delineated part on the basis of coordinates in the assigned grid. In some embodiments, the image data of the anatomical structure may be resampled using the assigned grid. Advantageous, a standardized addressing to the image data of the anatomical structure is provided, which may reduce the computational overhead of the machine learning, require fewer training data, etc.

First claim

The invention claimed is:

1. A system for preprocessing medical image data for machine learning, the system comprising:
an image data interface configured to access image data, the image data comprising an anatomical structure;
a memory comprising instruction data representing a set of instructions;
a processor configured to communicate with the image data interface and the memory and to execute the set of instructions, wherein the set of instructions, when executed by the processor, cause the processor to:
segment the anatomical structure in the image data to obtain a segmentation of the anatomical structure as a delineated part of the image data;
assign a grid to the delineated part of the image data, the grid representing a partitioning of an exterior and interior of the type of anatomical structure using grid lines, wherein said assigning comprises adapting the grid to fit the segmentation of the anatomical structure in the image data; and
provide a machine learning algorithm with an addressing to the image data in the delineated part on the basis of coordinates in the assigned grid.

2. The system according to claim 1 , wherein the set of instructions, when executed by the processor, cause the processor to resample the image data of the anatomical structure using the assigned grid to obtain resampled image data which is directly accessible at the coordinates of the assigned grid.

3. The system according to claim 1 , wherein the set of instructions, when executed by the processor, cause the processor to execute the machine learning algorithm using the image data of the anatomical structure as input.

4. The system according to claim 3 , wherein:
the image data of the anatomical structure represents training data for the machine learning algorithm, or
the image data of the anatomical structure represents new data to which the machine learning algorithm is applied.

5. The system according to claim 1 , wherein the set of instructions, when executed by the processor, cause the processor to assign the grid to the delineated part of the image data based on anatomical landmarks in the image data which are identified by said segmentation of the anatomical structure.

6. The system according to claim 5 , wherein the set of instructions, when executed by the processor, cause the processor to segment the image data using a segmentation model for the type of anatomical structure, wherein the segmentation model comprises labels corresponding to the anatomical landmarks.

7. The system according to claim 1 , further comprising a grid data interface to a database which comprises grid data defining the grid, and wherein the set of instructions, when executed by the processor, cause the processor to access the grid data from the database via the grid data interface.

8. The system according to claim 7 , wherein:
the database comprises grid data of different grids representing partitionings of an exterior and interior of different types of anatomical structures using grid lines; and
the set of instructions, when executed by the processor, cause the processor to access the grid data of the type of anatomical structure shown in the image data based on identification of the anatomical structure.

9. The system according to claim 7 , wherein:
the database comprises grid data of different grids, the different grids representing partitionings of an exterior and interior of the type of anatomical structure using grid lines for different medical applications; and
the set of instructions, when executed by the processor, cause the processor to:
obtain an identification of a current medical application, and
access the grid data of the type of anatomical structure shown in the image data and corresponding to the current medical application, based on the identification of the anatomical structure and the current medical application.

10. The system according to claim 9 , wherein the different grids representing partitionings of an exterior and interior of the type of anatomical structure for different medical applications differ at least locally in grid density.

11. The system according to claim 1 , further comprising a display interface to a display, and wherein the set of instructions, when executed by the processor, cause the processor to, via the display interface, establish a visualization of the assigned grid on the display.

12. The system according to claim 11 , wherein the visualization is an overlay of the assigned grid over the delineated part of the image data.

13. A workstation or imaging apparatus comprising the system according to claim 1 .

14. A computer-implemented method of preprocessing medical image data for machine learning, the method comprising:
accessing image data comprising an anatomical structure;
segmenting the anatomical structure in the image data to obtain a segmentation of the anatomical structure as a delineated part of the image data;
assigning a grid to the delineated part of the image data, the grid representing a partitioning of an exterior and interior of the type of anatomical structure using grid lines, said assigning comprising adapting the grid to fit the segmentation of the anatomical structure in the image data; and
providing a machine learning algorithm with an addressing to the image data in the delineated part on the basis of coordinates in the assigned grid.

15. The computer-implemented method according to claim 14 , further comprising:
accessing grid data from a database comprising grid data which defines the grid, wherein the grid data is of different grids representing partitionings of an exterior and interior of different types of anatomical structures using grid lines; and
wherein the accessed grid data is of the type of anatomical structure shown in the image data based on identification of the anatomical structure.

16. The computer-implemented method according to claim 15 , wherein grid lines of the different types of anatomical structures are for different medical applications, and further comprising:
obtaining an identification of a current medical application; and
accessing the grid data of the type of anatomical structure shown in the image data and corresponding to the current medical application, based on the identification of the anatomical structure and the current medical application.

17. The computer-implemented method according to claim 14 , further comprising establishing a visualization of the assigned grid on a display.

18. A non-transitory computer readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to:
access image data comprising an anatomical structure;
segment the anatomical structure in the image data to obtain a segmentation of the anatomical structure as a delineated part of the image data;
assign a grid to the delineated part of the image data, the grid representing a partitioning of an exterior and interior of the type of anatomical structure using grid lines, said assigning comprising adapting the grid to fit the segmentation of the anatomical structure in the image data; and
provide a machine learning algorithm with an addressing to the image data in the delineated part on the basis of coordinates in the assigned grid.

19. The non-transitory computer readable medium according to claim 18 , storing further instructions to:
access grid data from a database comprising grid data which defines the grid, wherein the grid data is of different grids representing partitionings of an exterior and interior

Assignees

• Koninklijke Philips Nv

Inventors

• Weese Rolf Jürgen
• Groth Alexandra
• Brosch Tom
• Peters Jochen

Classifications

• G06T7/11 Primary

  Region-based segmentation · CPC title

• G06N20/00

  Machine learning · CPC title

• G06T7/0012

  Biomedical image inspection · CPC title

• G06V10/469

  Contour-based spatial representations, e.g. vector-coding · CPC title

• G06V10/82

  using neural networks · CPC title