Feature-based composing for 3D MR angiography images

• US 8,265,354 B2
• Filed: 07/20/2005
• Issued: 09/11/2012
• Est. Priority Date: 08/24/2004
• Status: Active Grant

• Current Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Original Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Inventors: Zhang, Li, Sofka, Michal, Schäfer, Ulf
• Primary Examiner(s): Bali, Vikkram
• Application Number:
  US11/185,603
• Publication Number:
  US 20060052686A1
• Time in Patent Office: 2,610 Days
• Field of Search: 382/130, 382/131, 382/294, 382/298, 382/299, 382/266, 382/103, 382/173, 382/174, 382/190, 382/199, 382/201, 382/203, 382/205, 382/218, 382/219, 382/220, 382/256, 382/257, 382/258, 382/259, 382/278, 382/282, 382/283, 382/284
• US Class Current: 382/128
• CPC Class Codes: G06T 2200/04 involving 3D image data
• CPC Class Codes: G06T 2207/10088 Magnetic resonance imaging [MRI]
• CPC Class Codes: G06T 2207/30101 Blood vessel; Artery; Vein; Vascular
• CPC Class Codes: G06T 7/32 using correlation-based methods

Abstract

Multiple volumes that are to be aligned to form a single volume are processed. The system and method use an equalization step, a edge detection step and a correlation step to determine the overlapping positions between the first volume and the second volume of a volume pair having a maximum correlation value, and the best alignment of the first volume and the second volume of the volume pair is determined by the correlation value. A coarse correlation step using lower resolution volumes can be performed first followed by a fine correlation step using higher resolution images to save processing time. Initial preprocessing steps such as volume shearing can be performed. Equalization involves equalizing voxel size and edge detection can be performed using a Canny edge detector.

Claims

1. A method of aligning by a processor of a plurality of volumes from a patient, comprising:

reading by the processor of the plurality of volumes being Magnetic Resonance Angiography (MRA) volumes in DICOM format and correcting at least one of the plurality of volumes for shearing based on a tilt angle obtained from a DICOM header;

equalizing the plurality of volumes with the highest resolution of the plurality of volumes;

detecting edges in a plurality of 2D slices of a first volume of the plurality of volumes within a search range and a plurality of 2D slices of a second volume of the plurality of volumes within the search range to form a first and a second plurality of edge maps, respectively, the first and second edge maps having an overlap region wherein the overlap region depends upon the search range;

designating the first plurality of edge maps as a fixed image I_f with a location p=(p_x, p_y, p_z);

designating the second plurality of edge maps as a moving image I_t with a translation t=(t_x, t_y, t_z) with respect to the fixed image;

determining the search range based on a shift in a horizontal, vertical and depth direction of a parameter, the parameter determined from at least a relative table position and an image position; and

determining a correlation of the first plurality of edge maps to the second plurality of edge maps at a plurality of overlapping positions in the overlap region O containing the location p for all 2D slices in the overlap region of the first and the second volume to determine a best alignment of the first volume and the second volume.

2. The method as claimed in claim 1, further comprising:

detecting edges in the remaining plurality of volumes to form additional edge maps; and

determining the correlation of adjacent edge maps to determine the best alignment of the remaining plurality of volumes.

3. The method as claimed in claim 1, further comprising aligning the first volume and the second volume based on the correlation step to form a single image.

4. The method as claimed in claim 1, wherein the best alignment between the first volume and the second volume is found when the correlation of the plurality of overlapping positions is maximized.

5. The method as claimed in claim 1, wherein the correlating step is determined by maximizing the output of the formula:

6. The method as claimed in claim 1, wherein the step of equalizing equalizes the size of voxels in all the volumes.

7. The method as claimed in claim 1, wherein the step of detecting edges uses a Canny edge detector.

8. The method as claimed in claim 1, further comprising:

forming the first and second edge maps by down-sampling edge maps derived from volumes having full resolution, whereby the first and second edge maps represent a coarse resolution;

selecting an optimized alignment on the coarse resolution edge maps by maximizing coarse correlation values; and

determining an optimized alignment on the original resolution edge maps by maximizing fine correlation values around the optimized alignment on the coarse resolution.

9. The method as claimed in claim 1, wherein one or more voxels of a background in the first plurality of edge maps overlapping with one or more voxels of a background in the second plurality of edge maps do not contribute to the correlation.

10. A system for aligning a plurality of volumes from a patient, comprising:

means for reading the plurality of volumes being Magnetic Resonance Angiography (MRA) volumes in DICOM format and correcting at least one of the plurality of volumes for shearing based on a tilt angle obtained from a DICOM header;

means for equalizing the plurality of volumes with the highest resolution of the plurality of volumes;

means for detecting edges in a plurality of 2D slices of a first volume of the plurality of volumes within a search range and in a plurality of 2D slices of a second volume of the plurality of volumes within a search range to form a first and a second plurality of edge maps, respectively, the first and second edge maps having an overlap region, wherein the-overlap region of the first and the second volume depends upon the search range;

means for determining the search range based on a shift in a horizontal, vertical and depth direction of a parameter, the parameter determined from at least a relative table position and an image position; and

means for determining a correlation of the first plurality of edge maps to the second plurality of edge maps for all 2D slices in the overlap region of the first and the second volume to determine a best alignment of the first volume and the second volume.

11. The system as claimed in claim 10, further comprising:

means for detecting edges in the remaining plurality of volumes to form additional edge maps; and

means for determining the correlation of adjacent edge maps to determine the best alignment of the remaining plurality of volumes.

12. The system as claimed in claim 10, further comprising means for aligning the remaining volumes based on the correlation step to form a single image.

13. The system as claimed in claim 10, wherein the best alignment between the first volume and the second volume is found when the correlation of the plurality of overlapping positions is maximized.

14. The system as claimed in claim 10, wherein the means for determining the correlation maximizes the output of the formula:

15. The system as claimed in claim 10, wherein the means for equalizing equalizes the size of voxels in all the volumes.

16. The system as claimed in claim 10, wherein the means for detecting edges uses a Canny edge detector.

17. The system as claimed in claim 10, further comprising:

means for forming the first and second edge maps by down-sampling edge maps derived from volumes having full resolution, whereby the first and second edge maps represent a coarse resolution;

means for selecting an optimized alignment on the coarse resolution edge maps by maximizing coarse correlation values; and

means for determining an optimized alignment on the original resolution edge maps by maximizing fine correlation values.

18. The system as claimed in claim 10, wherein one or more voxels of a background in the first plurality of edge maps overlapping with one or more voxels of a background in the second plurality of edge maps do not contribute to the correlation.

19. A system for aligning a first volume of image data and a second volume of image data from a single patient, comprising:

a device for imaging a body part to form the first volume and the second volume;

a processor that reads the plurality of volumes being Magnetic Resonance Angiography (MRA) volumes in DICOM format and corrects at least one of the plurality of volumes for shearing based on a tilt angle obtained from a DICOM header, that equalizes data from the first volume and data from the second volume, that detects edges within a search range in the first volume to form a first plurality of edge maps of a plurality of 2D slices of the first volume, that detects edges within a search range in the second volume to form a second plurality of edge maps of a plurality of 2D slices of the second volume, the first and second edge maps having an overlap region, the overlap region being determined by a search range, based on shifts in a horizontal, vertical and depth direction of a parameter, the parameter determined from at least a relative table position and an image position, that performs determining a correlation between the first plurality of edge maps and the second plurality of edge maps at a plurality of overlapping positions in the search range between the first plurality of edge maps and the second plurality of edge maps, and that determines the best alignment of the first volume and the second volume based on the correlation; and

means for displaying a single image derived from the first and second volumes.

20. The system as claimed in claim 19, wherein a plurality of volumes from a single patient is processed and the best alignment of the plurality of volumes is determined.