Anterior commissure and posterior commissure segmentation system and method

• US 8,737,712 B2
• Filed: 09/14/2012
• Issued: 05/27/2014
• Est. Priority Date: 09/24/2008
• Status: Active Grant

• Current Assignee: Merative US LP
• Original Assignee: Merge Healthcare Incorporated (International Business Machines Corporation)
• Inventors: Geiger, Paul Arthur
• Primary Examiner(s): Gordadia, Shefali
• Application Number:
  US13/618,551
• Publication Number:
  US 20130163841A1
• Time in Patent Office: 620 Days
• Field of Search: 382/100, 382128-134
• US Class Current: 382/131
• CPC Class Codes: A61B 5/055 involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
• CPC Class Codes: G01R 33/546 Interface between the MR system and the user, e.g. for controlling the operation of the MR system or for the design of pulse sequences
• CPC Class Codes: G01R 33/5608 Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels image data processing or generation, in general G06T
• CPC Class Codes: G06T 2207/10072 Tomographic images
• CPC Class Codes: G06T 2207/30016 Brain
• CPC Class Codes: G06T 3/06 Topological mapping of higher dimensional structures onto lower dimensional surfaces
• CPC Class Codes: G06T 7/13 Edge detection
• CPC Class Codes: G06T 7/68 of symmetry
• CPC Class Codes: G06T 7/75 involving models

Abstract

A computer-implemented system and method of determining anterior commissure (AC) and posterior commissure (PC) points in a volumetric neuroradiological image. The method includes determining, by a computer, a mid-sagittal plane estimate to extract a mid-sagittal plane image from the volumetric neuroradiological image, and AC and PC point estimates in the mid-sagittal plane image. The method further includes determining, by the computer, a refined mid-sagittal plane estimate from the AC and PC point estimates to extract a refined mid-sagittal plane image, the AC point from the refined mid-sagittal plane image, and the PC point from the refined mid-sagittal plane image and the AC point.

Claims

1. A method of determining an anterior commissure (AC) point and a posterior commissure (PC) point in a volumetric image, the method comprising:

creating, by a computer, an edge mask of each of a plurality of axial image slices included in the volumetric image;

identifying, by the computer, a symmetry axis of each edge mask;

fitting, by the computer, a plane to the symmetry axes, the plane defining a mid-sagittal plane estimate;

extracting, by the computer, a mid-sagittal image from the volumetric image based on the mid-sagittal plane estimate; and

determining the AC point and the PC point in the mid-sagittal plane image.

2. The method of claim 1, wherein determining AC and PC points in the mid-sagittal plane image includes:

determining the AC point and the PC point in the mid-sagittal plane image;

determining a refined mid-sagittal plane estimate based on the AC point and the PC point;

extracting a refined mid-sagittal plane image from the volumetric image based on the refined mid-sagittal plane estimate; and

updating at least one of the AC point and the PC point based on the refined mid-sagittal plane image.

3. The method of claim 2, wherein updating at least one of the AC point and the PC point includes updating the PC point based on the refined mid-sagittal plane image and the AC point.

4. The method of claim 2, wherein determining the AC point and the PC point comprises applying an active appearance model to the mid-sagittal plane image.

5. The method of claim 4, wherein the active appearance model is one of a brainstem and third ventricle region.

6. The method of claim 2, wherein determining the refined mid-sagittal plane estimate comprises:

reformatting a plurality of axial image slices onto a plane which is perpendicular to the mid-sagittal plane estimate and which passes through the AC point and the PC point;

extracting an axial plane image;

cropping the axial plane image to a region including the AC point and the PC point and at least a portion of a brain ventricle extending therebetween;

creating an edge mask of the cropped axial plane image; and

identifying a symmetry axis of the cropped axial plane image.

7. The method of claim 6, wherein identifying the symmetry axis of the cropped axial plane image comprises:

(a) identifying a tentative symmetry axis, the tentative symmetry axis defining first and second halves of the edge mask;

(b) reflecting each pixel of the first half of the edge mask over the tentative symmetry axis;

(c) scoring one point for each pixel which lands on the second half of the edge mask; and

(d) adding each point into a total score for the tentative symmetry axis;

the method further comprising, iterating acts (a), (b), (c), and (d) for each of a range of tentative symmetry axes; and

identifying the tentative symmetry axis with the highest total score as the symmetry axis.

8. The method of claim 2, wherein updating at least one of the AC point and the PC point comprises:

projecting the AC point onto the refined mid-sagittal plane image;

identifying a brightness peak within a region surrounding the projected AC point; and

updating the AC point based on the brightness peak.

9. The method of claim 2, wherein updating at least one of the AC point and the PC point comprises:

projecting the PC point onto the refined mid-sagittal plane image;

creating an edge mask of the refined mid-sagittal plane image;

identifying a point X within a region surrounding the projected PC point, which lies on the edge mask, and for which a gradient of the refined mid-sagittal plane image is approximately parallel to a direction from the AC point to the point X;

obtaining a point Y by translating the point X along the direction from the AC point to the point X;

identifying an intensity peak in a direction perpendicular to the direction from the AC point to Y; and

updating the PC point based on the intensity peak.

10. The method of claim 1, further comprising equalizing brightness of the plurality of axial image slices prior to creating the edge mask.

11. The method of claim 1, wherein the plurality of axial image slices includes slices in a vertical range from a medulla to a top of a corpus calossum.

12. The method of claim 1, wherein identifying a symmetry axis of each edge mask includes:

(a) identifying a tentative symmetry axis, the tentative symmetry axis defining first and second halves of the edge mask;

(b) reflecting each pixel of the first half of the edge mask over the tentative symmetry axis;

(c) scoring one point for each pixel which lands on the second half of the edge mask; and

(d) adding each point into a total score for the tentative symmetry axis;

the method further comprising, iterating acts (a), (b), (c), and (d) for each of a range of tentative symmetry axes and identifying the tentative symmetry axis with the highest total score as the symmetry axis.

13. An image processing system configured to determine an anterior commissure (AC) point and a posterior commissure (PC) point in a volumetric image, the system comprising:

a first module configured to identify a mid-sagittal plane image included in the volumetric image using an AC point approximation and a PC point approximation;

a second module configured to identify the AC point based on the mid-sagittal plane image; and

a third module configured to identify the PC point based on the mid-sagittal plane image and the AC point.

14. The image processing system of claim 13, further comprising:

a first estimator configured to determine a mid-sagittal plane approximation based on the volumetric image; and

a second estimator configured to determine the AC point approximation and the PC point approximation based on the mid-sagittal plane approximation.

15. The image processing system of claim 13, wherein the first estimator is configured to determine the mid-sagittal plane approximation by:

creating an edge mask of each of a plurality of axial image slices included in the volumetric image;

identifying a symmetry axis of each edge mask;

fitting a plane to the symmetry axes;

determining, a mid-sagittal plane estimate based on fitting a plane to the symmetry axis; and

extracting the mid-sagittal plane approximation based on the mid-sagittal plane estimate.

16. The image processing claim 15, wherein the first estimator is further configured to equalize brightness of the plurality of axial image slices prior to creating the edge mask.

17. A non-transitory computer readable medium encoded with a plurality of processor-executable instructions for identifying an anterior commissure (AC) point and a posterior commissure (PC) point in a volumetric image, the instructions including instructions for:

determining a mid-sagittal plane estimate to extract a mid-sagittal plane image from the volumetric image;

estimating the AC point and the PC point in the mid-sagittal plane image;

determining a refined mid-sagittal plane based on the AC point and the PC point to extract a refined mid-sagittal plane image; and

updating at least one of the AC point and the PC point based on the refined mid-sagittal plane image.

18. The computer readable medium of claim 17, wherein the instructions for updating at least one of the AC point and the PC point includes instructions for updating the PC point based on the refined mid-sagittal plane image and the AC point.

19. The computer readable medium of claim 17, wherein the instructions for determining the mid-sagittal plane estimate includes instructions for:

creating an edge mask of each of a plurality of axial image slices included in the volumetric image;

identifying a symmetry axis of each edge mask;

fitting a plane to the symmetry axes, the plane defining the mid-sagittal plane estimate.