Contour extraction in multi-phase, multi-slice cardiac MRI studies by propagation of seed contours between images

• US 5,239,591 A
• Filed: 07/03/1991
• Issued: 08/24/1993
• Est. Priority Date: 07/03/1991
• Status: Expired due to Term

• Current Assignee: North American Philips Corp. (Koninklijke Philips N.V.)
• Original Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Inventors: Ranganath, Surendra
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): Cammarata, Michael
• Application Number:
  US07/725,144
• Time in Patent Office: 783 Days
• Field of Search: 382/6, 382/22, 382/19, 382/27, 382/48, 382/49, 382/9, 382/60, 395/141, 395/142, 395/924, 395/119, 128/653.1, 128/653.2, 324/309, 324/307
• US Class Current: 382/128
• CPC Class Codes: G01R 33/56 Image enhancement or correction, e.g. subtraction or averaging techniques , e.g. improvement of signal-to-noise ratio and resolution
• CPC Class Codes: G01R 33/563 of moving material, e.g. flow contrast angiography

Abstract

Contours are extracted from corresponding features the left ventricular blood pool-endocardium interfaces, of a plurality of images collected as a multi-phase multi-slice cardiac Magnetic Resonance Imaging (MRI) study which are accessible by a computer in response to user input to the computer of a seed contour identifying the contour feature to be extracted in an initial image at a middle slice position and a predetermined phase position, namely end of diastole. From this as the only contour inputted by the user, contours are extracted from each image by a sequence of automatic propagation of determinations of extracted, or final contours by forming from a final contour for an image, a seed contour for a not yet processed image which immediately adjoins in slice or phase position. For each image use is made of a seed-to-final propagation function in the form of an energy minimizing active contour function which has internal energy due to settable resistances to stretching and bending and image derived energy related to the gradient of the image. The final contour developed in the initial image is passed as a provisional contour to images which immediately adjoin in phase position where they are further propagated by an in-phase provisional-to-seed propagation function and are passed to images which immediately adjoin in slice position, where they are further propagated by an in-slice provisional-to-seed propagation function. After extraction of the contours, and possible automatic post-processing for correction thereof, these contours are used by a computational or visualization application, such as the computation of an ejection fraction.

Claims

1. A method of extracting final contours fit to a contour feature in a collection of images obtained by scanning an object undergoing a repetitive motion cycle with magnetic resonance imaging apparatus, each of said images representing a different combination of a slice position, selected from a plurality of successively spatially displaced slice positions in the object, and a phase position, selected from a plurality of successive phase positions in the motion cycle, said method comprising:

displaying on a display device an initial image from said collection of images, corresponding to a combination of a predetermined first phase position and a predetermined first slice position;

receiving and displaying in said initial image a seed contour input by a user, via an input device, proximate a contour feature in said initial image, to be extracted;

automatically propagating said seed contour with a seed-to-final propagation function to produce a final contour which is fit to said contour feature in said initial image;

said final contour being a provisional contour for a first pair of in-slice images at the first slice position and at phase positions which respectively immediately precede and immediately succeed said first phase position and for a first pair of in-phase images at the first phase position and at slice positions which respectively immediately precede and immediately succeed said first slice position;

automatically, without further action by the user, propagating said provisional contour for said first pair of in-slice images with an in-slice provisional-to-seed propagation function to respective seed contours for said first pair of in-slice images;

automatically, without further action by the user, propagating said provisional contour for said first pair of in-phase images with an in-phase provisional-to-seed propagation function to respective seed contours for said first pair of in-phase images; and

automatically propagating said seed contours for said first pairs of in-phase and in-slice images with said seed-to-final propagation function to respective final contours fit to said contour feature in said first pairs of in-phase and in-slice images;

wherein said seed-to-final propagation function comprises applying a plurality of iterations of an energy minimizing active contour function having an energy which is a function of energy derived from the image at the active contour, active contour stretching energy, and active contour bending energy.

2. The method as claimed in claim 1, wherein said final contours in said first pair of in-phase images are provisional contours for a second pair of in-phase images at the first slice position and at phase positions which respectively immediately precede a phase position of one of the first pair of in-phase images and immediately succeed a phase position of the other of said first pair of in-phase images, and further comprising:

automatically, without further action by the user, propagating said provisional contours for said second pair of in-phase images with said in-phase provisional-to-seed propagation function to respective seed contours for said second pair of in-phase images; and

automatically propagating said seed contours for second pair of in-phase images with said seed-to-final propagation function to respective final contours fit to said contour feature in said second pair of in-phase images.

3. The method as claimed in claim 2, wherein said final contours in said first pair of in-slice images are provisional contours for a second pair of in-slice images at the first phase position and at slice positions which respectively immediately precede a slice position of one of the first pair of in-slice images and immediately succeed a slice position of the other of said first pair of in-slice images, and further comprising:

automatically, without further action by the user, propagating said provisional contours for said second pair of in-slice images which said in-slice provisional-to-seed propagation function to respective seed contours for said second pair of in-slice images; and

automatically propagating said seed contours for second pair of in-slice images with said seed-to-final propagation function to respective final contours fit to said contour feature in said second pair of in-slice images.

4. The method as claimed in claim 2, wherein the final contour in one of said first pair of in-phase images is a provisional contour for a second pair of in-slice images at the same slice position as said one of said first pair of in-phase images and at phase positions which respectively immediately precede and immediately succeed said first phase position and the final contour in the other of said first pair of in-phase images as a provisional contour for a third pair of in-slice images at the same slice position as said other of said first pair of in-phase images and at phase positions which respectively immediately precede and immediately succeed said first phase position, and further comprising:

automatically, without further action by the user, propagating said provisional contours for said second and third pairs of in-slice images with said in-slice provisional-to-seed propagation function to respective seed contours for said second and third pairs of in-slice images; and

automatically propagating said seed contours for second and third pairs of in-slice images with seed-to-final propagation function to respective final contours fit to said contour feature in said second and third pairs of in-slice images.

5. The method as claimed in claim 3, wherein the final contour in one of said first pair of in-phase images is a provisional contour for a third pair of in-slice images at the same slice position as said one of said first pair of in-phase images and at phase positions which respectively immediately precede and immediately succeed said first phase position and the final contour in the other of said first pair of in-phase images is a provisional contour for a fourth pair of in-slice images at the same slice position as said other of said first pair of in-phase images and at phase positions which respectively immediately precede and immediately succeed said first phase position, and further comprising:

automatically, without further action by the user, propagating said provisional contours for said third and fourth pairs of in-slice images with said in-slice provisional-to-seed propagation function to respective seed contours for said third and fourth pairs of in-slice images; and

automatically propagating said seed contours for third and fourth pairs of in-slice images with said seed-to-final propagation function to respective final contours fit to said contour feature in said third and fourth pairs of in-slice images.

6. The method as claimed in claim 1, wherein said final contours in said first pair of in-slice images are provisional contours for a second pair of in-slice images at the first phase position and at slice positions which respectively immediately precede a slice position of one of the first pair of in-slice images and immediately succeed a

automatically, without further action by the user, propagating said provisional contours for said second pair of in-slice images with said in-slice provisional-to-seed propagation function to respective seed contours for said second pair of in-slice images;

and

automatically propagating said seed contours for second pair of in-slice images with said seed-to-final propagation function to respective final contours fit to said contour feature in said second pair of in-slice images.

7. The method as claimed in claim 6, wherein the final contour in one of said first pair of in-phase images is a provisional contour for a third pair of in-slice images at the same slice position as said one of said first pair of in-phase images and at phase positions which respectively immediately precede and immediately succeed said first phase position and the final contour in the other of said first pair of in-phase images is a provisional contour for a fourth pair of in-slice images at the same slice position as said other of said first pair of in-phase images and at phase positions which respectively immediately precede and immediately succeed said first phase position, and further comprising:

automatically, without further action by the user, propagating said provisional contours for said third and fourth pairs of in-slice images with said in-slice provisional-to-seed propagation function to respective seed contours for said third and fourth pairs of in-slice images; and

automatically propagating said seed contours for third and fourth pairs of in-slice images with said seed-to-final propagation function to respective final contours fit to said contour feature in said third and fourth pairs of in-slice images.

8. The method as claimed in claim 1, wherein the final contour in one of said first pair of in-phase images is a provisional contour for a second pair of in-slice images at the same slice position as said one of said first pair of in-phase images and at phase positions which respectively immediately precede and immediately succeed said first phase position and the final contour in the other of said first pair of in-phase images is a provisional contour for a third pair of in-slice images at the same slice position as said other of said first pair of in-phase images and at phase positions which respectively immediately precede and immediately succeed said first phase position, and further comprising:

automatically, without further action by the user, propagating said provisional contours for said second and third pairs of in-slice images with said in-slice provisional-to-seed propagation function to respective seed contours for said second and third pairs of in-slice images; and

automatically propagating said seed contours for second and third pairs of in-slice images with said seed-to-final propagation function to respective final contours fit to said contour feature in said second and third pairs of in-slice images.

9. The method as claimed in claim 1, wherein said in-phase provisions-to-seed propagation function comprises forming rough contours by matching intensity profiles in said initial image along normals to and proximate to said final contour in said initial image to intensity profiles in said first pair of in-phase images along said normals and proximate to said rough contours in said first pair of images, and applying a spatial smoothing function to said rough contours.

10. The method as claimed in claim 1, wherein said in-slice provisional-to-seed propagation function comprises thresholding said first pair of in-slice images to form for each, an in-slice thresholded image, determining extended edges in the respective in-slide thresholded images, and applying said energy minimizing active contour function with an energy derived from said extended edges.

11. The method as claimed in claim 9, wherein said in-slice provisional-to-seed propagation function comprises thresholding said first pair of in-slice images to form for each, an in-slice thresholded image, determining extended edges in the respective in-slice thresholded images, and applying said energy minimizing active contour function with an energy derived from said extended edges.

12. The method as claimed in claim 9, wherein said spatial smoothing function comprises applying said active contour function to said rough contours in the absence of energy derived from the image but with a predetermined limited number of iterations.

13. The method as claimed in claim 10, wherein said spatial smoothing function comprises applying said active contour function to said rough contours in the absence of energy derived from the image but with a predetermined limited number of iterations.

14. The method as claimed in claim 11, wherein said spatial smoothing function comprises applying said active contour function to said rough contours in the absence of energy derived from the image but with a predetermined limited number of iterations.

15. A method of determining a relative volume of a blood pool in a heart chamber of a patient at a pair of generally opposed phase positions in a beating cycle, comprising:

scanning a chest region of the patient with magnetic resonance imaging apparatus to obtain a collection of images of slices through the patient'"'"'s heart chamber, each of said images representing a different combination of a slice position, selected from a plurality of successively spatially displaced slice positions in the heart chamber, and a phase position, selected from a plurality of successive phase positions in the beating cycle;

displaying on a display device an initial image from said collection of images, corresponding to a combination of a predetermined one of said phase positions and a slice position at a middle one of said slice positions;

receiving and displaying in said initial image a seed contour input by a user, via an input device, proximate an inner wall of said chamber defining said blood pool in said initial image;

using said seed contour as the only user input contour proximate the inner wall of said chamber, successively extracting final contours defining said blood pool in said respective images by automatically, without further action by the user, propagating successive extracted final contours from respective images to seed contours for images which adjoin the respective images in slice or phase position, and automatically, without further action by the user, propagating said seed contours to final contours using a seed-to-final propagation function; and

using said extracted final contours for images in the first and second phase positions, computing said relative volume;

wherein said seed-to-final propagation function comprises applying a plurality of iterations of an energy minimizing active contour function having an energy which is a function of energy derived from the image at the active contour, active contour stretching energy, and active contour bending energy.

16. A method of obtaining final contours fit to a contour feature of an object undergoing a repetitive motion cycle, said method comprising:

scanning the object with magnetic resonance imaging apparatus to obtain a collection of images of the object, each of said images representing a different combination of a slice position, selected from a plurality of successively spatially displaced slice positions in the object, and a phase position, selected from a plurality of successive phase positions in the motion cycle;

displaying on a display device an initial image from said collection of images, corresponding to a combination of a predetermined first phase position and a predetermined first slice position;

receiving and displaying in said initial image a seed contour input by a user, via an input device, proximate a contour feature in said initial image, to be extracted;

automatically propagating said seed contour with a seed-to-final propagation function to produce a final contour which is fit to said contour feature in said initial image;

said final contour being a provisional contour for a first pair of in-slice images at the first slice position and at phase positions which respectively immediately precede and immediately succeed said first phase position and for a first pair of in-phase images at the first phase position and at slice positions which respectively immediately precede and immediately succeed said first slice position;

automatically, without further action by the user, propagating said provisional contour for said first pair of in-slice images with an in-slice provisional-to-seed propagation function to respective seed contours for said first pair of in-slice images;

automatically, without further action by the user, propagating said provisional contour for said first pair of in-phase images with an in-phase provisions-to-seed propagation function to respective seed contours for said first pair of in-phase images; and

automatically propagating said seed contours for said first pairs of in-phase and in-slice images with said seed-to-final propagation function to respective final contours fit to said contour feature in said first pairs of in-phase and in-slice images;

wherein said seed-to-final propagation function comprises applying a plurality of iterations of an energy minimizing active contour function having an energy which is a function of energy derived from the image at the active contour, active contour stretching energy, and active contour bending energy.