Method for determining the contour of an in vivo organ using multiple image frames of the organ

• US 5,734,739 A
• Filed: 08/26/1996
• Issued: 03/31/1998
• Est. Priority Date: 05/31/1994
• Status: Expired due to Term

• Current Assignee: University of Washington, Washington University In St Louis
• Original Assignee: University of Washington
• Inventors: Suri, Jasjit, Shao, Yong, Sheehan, Florence H., Haralick, Robert M.
• Primary Examiner(s): Couso, Jose L.
• Assistant Examiner(s): Do, Anh Hong
• Application Number:
  US08/704,016
• Time in Patent Office: 582 Days
• Field of Search: 382/128, 382/129, 382/130, 382/131, 382/132, 382/199, 378/98.2, 128/653.1, 128/659, 128/656, 250/369
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30048 Heart; Cardiac
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/155 involving morphological operators

Abstract

An automated method for evaluating image data taken over a sequence of image frames to determine a contour of a left ventricle (LV). The sequence of image frames are converted to digital data that identify a gray scale value for each pixel in each image frame. An initial LV region is estimated based upon training data parameters derived from processing manually drawn contours of the LV in the hearts of other individuals, corresponding to the sequence of image frames being analyzed. Prior to producing the initial estimate, the location of the aortic valve in the sequence of image frames for the patient is determined to serve as a reference. The data derived frown the contours of the LVs for the other hearts are rotated and translated relative to this reference to insure that the training data information corresponds to the position and orientation of the LV being analyzed. In addition, the effects of tissue mass comprising a diaphragm adjacent to the inferior portion of the LV are compensated by masking out the diaphragm. Uncertainty of the contour of the LV in this portion due to the presence of the diaphragm is thereby minimized. The resulting automatically determined contours of each image frame can be displayed to enable a physician to more readily diagnose physiological defects of the heart.

Claims

1. A method for automatically determining a contour of an internal organ in a patient based upon digital image data of a region in which the organ is disposed, said image data representing a sequence of image frames of the region made over an interval of time during which a wall of the organ has completed at least one cycle of movement, the wall of the organ undergoing repetitive cycles of movement as part of the organ'"'"'s physiological function, each image frame of said sequence of image frames comprising a plurality of pixels, said method comprising the steps of:

(a) developing an initial estimate of a region bounded by the contour of the organ in each image frame of the sequence, by comparing the image data with parameters derived from training data that were previously produced through manual evaluation of a plurality of image frames over a corresponding cycle for each of a plurality of corresponding organs in other individuals, said parameters that are derived from the training data being used to assign classes to the pixels comprising each of the image frames of the sequence, said classes indicating pixels that are most likely inside the contour of the organ in each image frame; and

(b) determining a border of the organ in a portion of the contour adjacent to a tissue mass lying outside the organ, by masking out the tissue mass, thereby refining the initial estimate of the region bounded by the contour of the organ so that the border of the organ is more accurately defined in said portion of the contour.

2. The method of claim 1, wherein an edge detector is used for detecting the border of the organ, producing an edge boundary that is used to determine the border in said portion of the contour.

3. The method of claim 2, further comprising the steps of:

(a) applying the training data to determine a masked region adjacent to and outside said portion of the contour; and

(b) combining the masked region with the edge boundary to refine the initial estimate of the region and thereby to produce a more accurate contour of the organ at said portion of the contour.

4. The method of claim 3, wherein the masked region is determined using a morphological process.

5. The method of claim 4, wherein the morphological process employs:

(a) one of a morphological opening and closing process;

(b) an automatic thresholding process that applies a threshold for determining a state of pixels in the masked region; and

(c) the training data.

6. The method of claim 2, further comprising the step of using dynamic programming to link edge row and column components together to form the edge boundary.

7. The method of claim 1, further comprising the step of more accurately defining the border for each image frame, by applying global shape constraints that relate physical shape attributes of the organ to the contour bounding the region.

8. The method of claim 1, wherein the step of developing an initial estimate of the region bounded by the contour of the organ in each image frame of the sequence comprises the steps of:

(a) estimating a reference location associated with an anatomical feature of the patient'"'"'s organ using the image frames of the sequence; and

(b) aligning a reference location associated with a corresponding anatomical feature in the training data with the reference location associated with the anatomical feature of the patient'"'"'s organ, prior to assigning classes to the pixels, to ensure that the classes assigned to the pixels as a function of the training data are not in error due to differences in the disposition of the organ in the patient and the training data.

9. The method of claim 8, wherein the step of estimating the reference location associated with the anatomical feature comprises the steps of:

(a) creating a synthetic image having a relatively maximum contrast, by selecting a brightest pixel at each pixel position from among each of the image frames comprising the sequence, for application to each corresponding pixel comprising the synthetic image;

(b) as a function of an extent of the anatomical feature in the training data, estimating a limited region for the anatomical feature in the synthetic image;

(c) estimating a boundary for the anatomical feature within the limited region; and

(d) as a function of the boundary estimated for the anatomical feature, estimating the reference location related to the anatomical feature.

10. The method of claim 9, wherein the step of estimating the boundary comprises the step of estimating a center of the anatomical feature.

11. The method of claim 9, wherein the step of estimating the boundary comprises the steps of projecting a gray scale profile along a predetermined angle, and as a function of the gray scale profile, determining a limited region for the contour of the anatomical feature.

12. The method of claim 1, wherein an edge detector is used for detecting the border of the tissue mass adjacent to the organ, producing an edge boundary that is used to determine a border segment between the organ and the tissue mass, said border segment representing an anatomic feature of the patient'"'"'s organ.

13. The method of claim 12, further comprising the steps of:

(a) applying the training data to estimate a limited region adjacent the border segment;

(b) estimating a boundary for the tissue mass as a function of the limited region; and

(c) as a function of the boundary estimated for the tissue mass, estimating a disposition of the anatomic feature.

14. The method of claim 1, wherein the method is embodied in a computer program that is stored on a memory media and adapted to be executed by a computer to implement the steps of the method.

15. A method for automatically determining a contour of an internal organ in a patient based upon digital image data of a region in which the organ is disposed, said image data representing a sequence of image frames of the region made over an interval of time during which a wall of the organ has completed at least one cycle of movement, the wall of the organ undergoing repetitive cycles of movement as part of the organ'"'"'s physiological function, each image frame of said sequence of image frames comprising a plurality of pixels, said method comprising the step of:

(a) aligning a reference location associated with an anatomical feature of the organs represented by training data with a reference location associated with a corresponding anatomical feature of the organ in the patient, where the training data were previously produced from a plurality of image frames over a corresponding cycle for each of a plurality of corresponding organs in other individuals; and

(b) developing an estimate of a region bounded by the contour of the organ in each image frame of the sequence, based upon a comparison of the image data with parameters derived through manual evaluation of the training data, said parameters derived from the training data being used to assign classes to the pixels comprising each of the image frames of the sequence, said classes indicating pixels that are most likely inside the contour of the organ in each image frame.

16. The method of claim 15, wherein the reference location is disposed at one end of the anatomical feature.

17. The method of claim 15, wherein the step of estimating comprises the step of producing a synthetic image comprising a plurality of pixels, each of said plurality of pixels of the synthetic image comprising a brightest pixel from each corresponding pixel of the sequence of image frames.

18. The method of claim 17, further comprising the step of estimating a region that contains said anatomical feature within the synthetic image, as a function of the training data.

19. The method of claim 18, further comprising the step of estimating contour boundary regions within the feature region, said boundary regions encompassing substantial portions of the contour of the anatomical feature.

20. The method of claim 19, further comprising the step of employing the training data to estimate the contour boundary regions.

21. The method of claim 19, wherein a dynamic programming search is employed to estimate a contour of the anatomical feature.

22. The method of claim 19, further comprising the step of removing anti-parallel line pairs, said anti-parallel line pairs comprising non-anatomic elements within the digital image data.

23. The method of claim 19, further comprising the step of estimating a middle curve of the anatomic feature.

24. The method of claim 19, further comprising the step of projecting gray scale levels within the feature region at a predefined angle.

25. The method of claim 15, wherein the organ comprises a left ventricle and the anatomical feature comprises an aortic valve.

26. The method of claim 15, wherein the method is embodied in a computer program that is stored on a memory media and adapted to be executed by a computer in order to implement the steps of the method.