METHOD FOR SAMPLING VOLUME DATA OF AN OBJECT IN AN IMAGING DEVICE

• US 20110170756A1
• Filed: 01/08/2010
• Published: 07/14/2011
• Est. Priority Date: 01/08/2010
• Status: Active Grant

• Current Assignee: Siemens Healthineers AG (Siemens AG)
• Original Assignee: Siemens AG
• Inventors: SCHNEIDER, Robert
• Granted Patent: US 8,682,053 B2
• Time in Patent Office: Days
• US Class Current: 382/131
• CPC Class Codes: G06T 15/08 Volume rendering

Abstract

A method for sampling volume data of an object in an imaging device is disclosed. In at least one embodiment, the method includes creating a voxel dataset of a first volume associated with the object by way of a scanning device and saving the voxel dataset in a main memory; defining a grid of sampling positions by casting a plurality of rays through the first volume and choosing a number of points along the respective ray; choosing a sampling access pattern from a plurality of sampling access patterns defining a sequence of the sampling positions; selecting an interpolation scheme mapping the sampling access pattern to a voxel access pattern defining a sequence of the voxels of the voxel dataset; according to the voxel access pattern, successively loading memory lines of the main memory comprising the data associated with the respective voxel into a cache memory in case the respective memory line is not already present in the cache memory; and processing the data associated with the respective voxel, the cache memory access being optimized for increasing rendering time performance. To this end in at least one embodiment, during the choosing of sampling access, patterns a block of sampling positions having a non-vanishing intersection with the grid is defined wherein the group of sampling access patterns is restricted to the set of recursive subdivisions of the block.

Claims

1. A method for sampling volume data of an object in an imaging device, the method comprising:

creating a voxel dataset of a first volume associated with the object by way of a scanning device and saving the created voxel dataset in a main memory;

defining a grid of sampling positions by casting a plurality of rays through the first volume and choosing a number of points along a respective ray;

choosing a sampling access pattern from a plurality of sampling access patterns defining a sequence of the defined grid of sampling positions;

selecting an interpolation scheme mapping the chosen sampling access pattern to a voxel access pattern defining a sequence of the voxels of the voxel dataset;

according to the voxel access pattern, successively loading memory lines of the main memory including the data associated with the respective voxel into a cache memory in case a respective memory line is not already present in the cache memory; and

processing the data associated with the respective voxel,

wherein during said choosing of sampling access patterns, a block of sampling positions having a non-vanishing intersection with said grid is defined and

wherein said group of sampling access patterns is restricted to the set of recursive subdivisions of the block.

2. The method of claim 1 wherein said group of sampling access patterns is further restricted to those satisfying the property that

when restricting the sampling access pattern to an arbitrary ray, the sampling positions along this ray are accessed monotonically.

3. The method of claim 2 wherein the size of the block in each spatial direction is an exponential of 2.

4. The method of claim 3 wherein said plurality of rays are parallel with respect to each other.

5. The method of claim 4 wherein said number of points along a respective ray is chosen equidistantly.

6. The method of claim 1 wherein said choosing of a sampling access pattern further comprises:

recursively subdividing the block into sub-blocks; and

choosing a sub-block from a plurality of sub-blocks respectively resulting from subdivisions in several spatial dimensions for a next step of said recursive subdivision.

7. The method of claim 6, wherein, during each step of the subdivision, the number of memory lines loaded for each of the respective resulting sub-blocks is approximated and the sub-block having a relatively least number of memory lines loaded is chosen for a next step of said recursive subdivision.

8. The method of claim 7, wherein the number of memory lines loaded is approximated by

determining a convex hull of an affine transformation of the block or sub-block into voxel space, and

calculating an area measure of a projection to a selected plane of a second volume associated with the convex hull.

9. The method of claim 8, wherein the voxel dataset of the first volume is created by way of the scanning device serially scanning tomographic cross sections of said first volume and wherein the selected plane is normal to the plane of the tomographic cross sections.

10. The method of claim 8, wherein the second volume is determined extending the convex hull by all spatial points having a distance to the convex hull that is relatively smaller than a selected distance value.

11. The method of claim 1, wherein sampling positions outside the object are removed from the grid of sampling positions.

12. The method of claim 1, wherein a plurality of blocks of sampling positions covering all of the grid is defined.

13. The method of claim 6, wherein

a parallelepiped defined by an affine transformation of the block or sub-block into voxel space is determined and

the sub-block resulting from the subdivision in the spatial dimension corresponding to a relatively largest edge of the parallelepiped is chosen for a next step of said recursive subdivision.

14. The method of claim 1, wherein the group of sampling access patterns is further restricted to those not including subdivisions in a selected spatial dimension after a selected number of steps of said recursive subdivision.

15. The method of claim 1, wherein the group of sampling access patterns is further restricted to those including repeating sequences of subdivisions in certain spatial dimensions.

16. The method of claim 12, wherein the processing is not performed for a sub-block having all sampling positions outside the object.

17. The method of claim 1, wherein processing the data associated with the respective voxel for creation of a two-dimensional image comprises:

linearly saving the data to a memory area according to the sampling access pattern, and

after having processed all data associated with respective rays of a plurality of pixels, permutating the saved data according to a swizzling pattern for creation of said two-dimensional image.

18. The method of claim 17, wherein the swizzling pattern is determined by a two-dimensional reduction of the sampling access pattern.

21. A computer readable medium including program segments for, when executed on a computer device, causing the computer device to implement the method of claim 1.

19. An imaging device, comprising:

means for creating a voxel dataset of a first volume associated with an object by way of a scanning device and saving the created voxel dataset in a main memory;

means for defining a grid of sampling positions by casting a plurality of rays through the first volume and choosing a number of points along a respective ray;

means for choosing a sampling access pattern from a plurality of sampling access patterns defining a sequence of the defined grid of sampling positions;

means for selecting an interpolation scheme mapping the chosen sampling access pattern to a voxel access pattern defining a sequence of the voxels of the voxel dataset;

means for, according to the voxel access pattern, successively loading memory lines of the main memory including the data associated with the respective voxel into a cache memory in case a respective memory line is not already present in the cache memory; and

means for processing the data associated with the respective voxel,

wherein during said choosing of sampling access patterns, a block of sampling positions having a non-vanishing intersection with said grid is defined and

wherein said group of sampling access patterns is restricted to the set of recursive subdivisions of the block.

20. An medical imaging appliance, comprising:

means for creating a voxel dataset of a first volume associated with an object by way of a scanning device and saving the created voxel dataset in a main memory;

means for defining a grid of sampling positions by casting a plurality of rays through the first volume and choosing a number of points along a respective ray;

means for choosing a sampling access pattern from a plurality of sampling access patterns defining a sequence of the defined grid of sampling positions;

means for selecting an interpolation scheme mapping the chosen sampling access pattern to a voxel access pattern defining a sequence of the voxels of the voxel dataset;

means for, according to the voxel access pattern, successively loading memory lines of the main memory including the data associated with the respective voxel into a cache memory in case a respective memory line is not already present in the cache memory; and

means for processing the data associated with the respective voxel,

wherein during said choosing of sampling access patterns, a block of sampling positions having a non-vanishing intersection with said grid is defined and

wherein said group of sampling access patterns is restricted to the set of recursive subdivisions of the block.