Automated method for digital image quantitation

• US 5,360,006 A
• Filed: 12/29/1993
• Issued: 11/01/1994
• Est. Priority Date: 06/12/1990
• Status: Expired due to Term

• Current Assignee: University of Florida Research Foundation, Incorporated (State University System of Florida)
• Original Assignee: University of Florida Research Foundation, Incorporated (State University System of Florida)
• Inventors: Gibby, Gordon L., Wilson, David C., Geiser, Edward A.
• Primary Examiner(s): JAWORSKI, FRANCIS J
• Application Number:
  US08/174,763
• Time in Patent Office: 307 Days
• Field of Search: 128/653.1, 128/650.07, 128/660.07, 128/661.04, 128/695, 364/413.25
• US Class Current: 600/425
• CPC Class Codes: G01S 7/52036 using analysis of echo signal for target characterisation

Abstract

A method is provided for automatically determining quantitative characteristics of organ images, especially echocardiographic images. A diagnostic image is obtained and then a center point for the approximately curvilinear bounded organ is determined by repeatedly filtering the image with a set of circular arc filters until a maximum value for the filter set is obtained, whereupon parameters of cardiac wall motion, wall thickness and area change fraction may be determined to detect ischemic or other organ abnormality.

Claims

1. A method for quantitatively analyzing diagnostic images of approximately curvilinear bounded organs, including the steps of:

a. generating a diagnostic image of an approximately curvilinear bounded organ;

b. determining a center-point for the approximately curvilinear bounded organ depicted in the diagnostic image by repeatedly filtering the diagnostic image with a set of circular arc filters until a maximum value for the set of filters is obtained;

c. indicating the position of the center-point in connection with the diagnostic image.

2. The method of claim 1, wherein the step of filtering the diagnostic image includes the step of:

computing for a selected set of pixels x_i,y_j of the diagnostic image the maximum of a set of numbers s_k -s_k-1, where s_k represents the sum of pixel values on a selected set of radii uniformly distributed along a semicircular arc of radius k.

3. The method of claim 1, further including the steps of:

a. determining a best-fit elliptical arc model of a border of the approximately curvilinear bounded organ, the elliptical model being centered at the center-point;

b. indicating the position of the elliptical arc model in connection with the diagnostic image.

4. The method of claim 3, wherein the step of determining a best-fit elliptical arc model includes the step of:

a. generating a one-dimensional histogram of radial pixel values for each of a set of sectors centered on the center-point.

5. The method of claim 4, wherein the step of determining a best-fit elliptical arc model includes the steps of:

a. generating a first derivative histogram for each sector from each corresponding one-dimensional histogram of radial pixel values;

b. determining at least one peak value of the first derivative histogram for each sector, representing a probable location of the border of the approximately curvilinear bounded organ within the sector.

17. The method of claim 1, wherein the step of filtering the diagnostic image includes the steps of:

a. for a selected first set of pixels x_i,y_j of the diagnostic image, computing the maximum of a set of numbers s_k -s_k-1 +s_k -s_k-2, where s_k represents the sum of pixel values on a selected set of radii uniformly distributed along a semicircular arc of radius k, thereby determining a first approximation of the center-point;

b. for a selected second set of pixels x_k,y_j of the diagnostic image within a first distance from the first approximated center-point, computing the maximum of a set of numbers s_k -s_k-1 +s_k -s_k-2, thereby determining a second approximation of the center-point;

c. for a selected third set of pixels x_m,y_n of the diagnostic image within a second distance from the second approximated center-point, the second distance being less than the first distance, computing the maximum of a set of numbers s_k -s_k-1, thereby determining a third approximation of the center-point.

6. A method for determining the motion of a border of an approximately curvilinear bounded organ, including the steps of:

a. generating a first and second diagnostic image of an approximately curvilinear bounded organ, the diagnostic images being generated at different times;

b. determining a center-point of the approximately curvilinear bounded organ for each of the first and second diagnostic images by repeatedly filtering each of the diagnostic images with a set of circular arc filters until a maximum value for the corresponding set of filters is obtained;

c. generating a first and second one-dimensional histogram of radial pixel values for each of a set of sectors centered on the center-point of each of the corresponding first and second diagnostic images;

d. generating a first and second first derivative histogram for each sector from each corresponding first and second one-dimensional histograms of radial pixel values;

e. determining the maximum cross-correlation of the first and second first derivative histograms for each sector;

f. indicating the maximum cross-correlation in each sector as a measure of the change in position in each sector of the border of the approximately curvilinear bounded organ between the first and second diagnostic images.

7. The method of claim 6, including the further steps of:

a. determining the average velocity of the change of position of the border for each sector, by dividing the measure of change of position for each sector by the time difference between the generation of the first and second diagnostic images;

b. indicating the average velocity so determined for each sector.

8. A method for determining the lack of a sufficient image signal across the interventricular septum of an echocardiogram, comprising the steps of:

a. generating a digital ultrasonic image of a heart;

b. determining a center-point for the left ventricle of the heart depicted in the diagnostic image by repeatedly filtering the diagnostic image with a set of circular arc filters until a maximum value for the set of filters is obtained;

c. determining the minimum average grey level of an image region in the left ventricle of the heart positioned around the center-point, thereby determining a background signal level;

d. determining the maximum average grey level for a set of image regions positioned along a line from the center-point of the left ventricle to at least the interventricular septum, thereby determining a peak signal level;

e. comparing the background signal level to the peak signal level;

f. if the maximum ratio of the peak signal level to the background signal level is less than about two, then indicating that the image signal across the interventricular septum of the echocardiogram is insufficient.

9. A method for determining the occurrence of a flattened interventricular septum from a diagnostic image of a heart, comprising the steps of:

a. generating a diagnostic image of a heart;

b. determining a center-point for the left ventricle of the heart depicted in the diagnostic image by repeatedly filtering the diagnostic image with a set of circular arc filters until a maximum value for the set of filters is obtained;

c. generating a best-fit to the endocardial and epicardial borders of the interventricular septum of a pair of straight line filters oriented perpendicular to a line from the center-point to at least the interventricular septum;

d. generating a best-fit to the endocardial and epicardial borders of the interventricular septum of a pair of circular arc filters centered on the center-point and encompassing at least the interventricular septum;

e. determining a figure of merit N_c for the pair of circular arc filters and a figure of merit N_s for the pair of straight line filters;

f. if N_s is greater than N_c, then indicating that the interventricular septum is flattened.

10. A method for quantitatively analyzing diagnostic images of the heart, including the steps of:

a. generating a diagnostic image of the heart;

b. determining a first tentative center-point for and a region of search for the posterior epicardial border of the heart depicted in the diagnostic image by repeatedly filtering the diagnostic image with a set of circular arc filters until a maximum value for the set of filters is obtained;

c. determining a second tentative center-point for and a region of search for the anterior endocardial and epicardial borders of the heart depicted in the diagnostic image by repeatedly filtering the diagnostic image with a set of coupled circular arc filters until a maximum value for the set of coupled filters is obtained;

d. determining a final center-point for the heart depicted in the diagnostic image as midway between the anterior epicardial border and the posterior epicardial border along a line through the first and second tentative center-points;

e. indicating the position of the final center-point in connection with the diagnostic image.

11. The method of claim 10, wherein the step of filtering the diagnostic image in determining the first tentative center-point includes the step of:

a. computing for a selected set of pixels x_i,y_j of the diagnostic image the maximum of a set of numbers s_k -s_k-1, where s_k represents the sum of pixel values on a selected set of radii uniformly distributed along a semicircular arc of radius k.

12. The method of claim 10, wherein the step of filtering the diagnostic image in determining the second tentative center-point and the region of search for the anterior endocardial border includes the step of:

a. computing, for a selected set of pixels x_i,y_j of the diagnostic image within a selected distance from the first tentative center-point, the maximum of a set of numbers s_k -s_k-1, where s_k represents the sum of pixel values on a selected set of radii uniformly distributed along an arc of radius k.

13. The method of claim 10, wherein the step of filtering the diagnostic image in determining the second tentative center-point and the region of search for the anterior epicardial border includes the step of:

a. computing, for a selected set of pixels x_i,y_j of the diagnostic image within a selected distance from the first tentative center-point, the maximum of a set of numbers max (s_k -s_k+1, 2·

s_k -s_k-1 -s_k+1)+t_k -t_k+1, where each s_k is the sum of pixel values on a first set of radii uniformly distributed along an arc of radius k, and each t_k is the sum of pixel values on a second set of radii uniformly distributed along the arc of radius k.

14. The method of claim 13, further including the steps of:

a. determining a best-fit elliptical arc model of the epicardial and anterior endocardial borders of the heart, the elliptical model being centered at the final center-point;

b. indicating the position of the elliptical arc model in connection with the diagnostic image of the heart.

15. The method of claim 14, wherein the step of determining a best-fit elliptical arc model includes the step of:

a. generating a one-dimensional histogram of radial pixel values for each of a set of sectors centered on the final center-point.

16. The method of claim 15, wherein the step of determining a best-fit elliptical arc model includes the steps of:

a. generating a first derivative histogram for each sector from each corresponding one-dimensional histogram of radial pixel values;

b. determining at least two peak values of the first derivative histogram for each sector, representing the probable locations of the epicardial and endocardial borders of the heart within the sector.

18. The method of claim 10, wherein the step of filtering the diagnostic image in determining the first tentative center-point includes the step of:

a. for a selected first set of pixels x_i,y_j of the diagnostic image, computing the maximum of a set of numbers s_k -s_k-1 +s_k -s_k-2, where s_k represents the sum of pixel values on a selected set of radii uniformly distributed along a semicircular arc of radius k, thereby determining a first approximation of the tentative center-point;

b. for a selected second set of pixels x_k,y_j of the diagnostic image within a first distance from the first approximated center-point, computing the maximum of a set of numbers s_k -s_k-1 +s_k -s_k-2, thereby determining a second approximation of the tentative center-point;

c. for a selected third set of pixels x_m,y_n of the diagnostic image within a second distance from the second approximated center-point, the second distance being less than the first distance, computing the maximum of a set of numbers s_k -s_k-1, thereby determining a third approximation of the tentative center-point.

19. The method of claim 10, wherein the step of filtering the diagnostic image in determining the second tentative center-point and the region of search for the anterior epicardial border includes the steps of:

a. for a selected set of pixels of the diagnostic image within a selected distance from the first tentative center-point, computing a distribution D(y) as the sum of the pixel values in a set of circular arcs centered at the first tentative center-point at a distance y;

b. for each value y in the distribution D(y), computing a first weighting function epiW(y) as a measure of the probability that pixels at location y are part of the epicardial border, and computing a second weighting function endoW(y) as a measure of the probability that pixels at location y are part of the endocardial border;

c. for a selected set of pixels x_k,y_j of the diagnostic image within a selected distance from the first tentative center-point, computing the maximum of a set of numbers;

space="preserve" listing-type="equation">V(x,y,k,k+c)=(s.sub.k -s.sub.k+c)×

epiW(y.sub.1)×

(k-(k+c))×

(s.sub.k+c -s.sub.k)×

endoW(y.sub.2)

where s_k represents the sum of pixel values on a selected set of radii uniformly distributed along a semicircular arc of radius k, y₁ <

y₂, y₁ =y-k, y₂ =y-(k +c), and c is a constant.

20. The method of claim 19, wherein the first and second weighting functions are initially computed to have values of 0 or 1, and further including the steps of:

a. computing a second distribution E(y) as the average [D(y-n)+D(y-n-1) . . . D(y+n-1)+D(y+n)]/n, where n is a selected distance from the first tentative center-point;

b. assigning a value of 2 to a weighting function at location y if the weighting function is initially assigned a value of 1, and if;

space="preserve" listing-type="equation">E(y-n)<

(2×

E.sub.min +E.sub.max)/3

where E_min is the minimum value of the second distribution E(y), and E_max is the maximum value of the second distribution E(y).

21. A method for quantitative blood pool classification from diagnostic images of the heart, including the steps of:

a. generating a diagnostic image of the heart;

b. determining a first tentative center-point for and a region of search for the posterior epicardial border of the heart depicted in the diagnostic image by repeatedly filtering the diagnostic image with a set of circular arc filters until a maximum value for the set of filters is obtained;

c. determining a second tentative center-point for and a region of search for the anterior endocardial and epicardial borders of the heart depicted in the diagnostic image by repeatedly filtering the diagnostic image with a set of coupled circular arc filters until a maximum value for the set of coupled filters is obtained;

d. determining a final center-point for the heart depicted in the diagnostic image as midway between the anterior epicardial border and the posterior epicardial border along a line through the first and second tentative center-points;

e. determining a best-fit elliptical arc model of the border of the heart, the elliptical arc model being centered at the center-point;

f. classifying pixels within the elliptical arc model whose values do not exceed a threshold value as being part of the blood pool;

g. summing the pixels classified as being part of the blood pool to determine a blood pool value, BP.

22. The method of claim 21, wherein the step of classifying pixels comprises the further steps of

a. determining a normalized radial distribution of a plurality of sets of pixels, centered on the final center-point;

b. determining an average value for the pixels in the plurality of sets of pixels;

c. assigning each set of pixels having a normalized value less than a selected percentage of the average value as being part of the blood pool.

23. The method of claim 21, wherein the step of classifying pixels comprises the further steps of:

a. determining an average value for the pixels in at least one defined region within the elliptical arc model;

b. assigning each pixel within at least one defined region and having a value less than a selected percentage of the average value as being part of the blood pool of such at least one region.

24. The method of claim 22, wherein a plurality of regions within the elliptical arc model are defined as equiangular regions centered on the final center-point.

25. The method of claim 21, comprising the further steps of:

a. determining the blood pool value at end diastole, BP_ED ;

b. determining the blood pool value at end systole, BP_ES ;

c. computing the area change fraction ACF for the region defined by the elliptical arc model as;

space="preserve" listing-type="equation">ACF=100%×

(BP.sub.ED -BP.sub.ES)/BP.sub.ED.