Adaptive processing of contrast enhanced ultrasonic diagnostic images
A method is described for automatically and accurately defining tissue borders within ultrasound imaging. A preferred embodiment described provides for accurately delineating endocardial and epicardial tissue borders during ultrasound imaging of the heart using contrast agents. The inventive method requires first defining a set of threshold values corresponding to various imaging depths and correlated within the image. The invention further uses the set of threshold values and a maximization function to accurately identify the tissue boundaries.
1. A method of defining a tissue border in a medical ultrasound image, comprising: acquiring a 2-D planar ultrasound image; identifying at least one key point or landmark within the ultrasound image; locating and identifying a center scanline in the ultrasound image relative the key point or benchmark; determining an array of threshold values correlated to the center scanline as a function of image depth; filling a trace to a tissue border related to the center scanline and the array of threshold values; and displaying an ultrasonic image with a fitted trace displayed at the tissue border.
2. The method of claim 1 , wherein the step of locating comprises matching a geometric template corresponding to the geometric relationship of the at least one key point or benchmark in the ultrasound image to the scanline.
3. The method of claim 1 , wherein the step of identifying includes that the at least one key point or benchmark is input automatically.
4. The method of claim 1 , wherein the step of identifying includes that the at least one key point or benchmark is input by a user.
5. The method of claim 1 , wherein the step of identifying includes that the number of key points or benchmarks is three.
6. The method of claim 1 , wherein the medical ultrasound images may be three-dimensional (3-D).
7. The method of claim 1 . wherein the step of identifying includes identifying the mitral valve in the image, and the step of fitting includes determining a series of chords out from the scanline to the tissue border, which chords are parallel to the mitral valve plane and occur at regular intervals.
8. The method of claim 7 , wherein at the origin of each chord, small neighborhoods of pixel intensity values are summed, and a mean is determined for each chord.
9. The method of claim 8 , wherein a threshold value is determined for each chord as a percentage of the mean value found for each chord.
10. The method of claim 7 , wherein the medical ultrasound image is contrast agent-enhanced.
11. The method of claim 7 , further including a step of evaluating each pixel value of a particular chord, beginning with the chord pixel value at the center scanline, and if the intensity value for each pixel value is found to be below the threshold for N consecutive pixels, qualifying an area comprising the N values as an edge.
12. The method of claim 11 , wherein the step of evaluating requires that a polynomial of degree N be fit through all of the threshold-determined points of a particular edge, calculating a mean distance between the polynomial fit and actual threshold-determined points and eliminating outlier points which are further than a factor F times the mean distance.
13. The method of claim 12 , further including a step for determining the best polynomial approximation to the real edges through the use of a maximization function.
14. The method of claim 12 , wherein the maximization function uses the mean intensity along the polynomial, the number of threshold points used to approximate the polynomial and the chi-squared error of the polynomial approximation to the threshold points.
15. The method of claim 14 where the chi-squared approximation may be substituted for an arbitrary “goodness-of-fit” estimate.
16. The method of claim 14 where the maximization function can contain any number of parameters.
17. The method of claim 1 , wherein the ultrasound image is a cardiac image and the three key points define the mitral valve and apex of the LV cavity.
18. The method of claim 17 , wherein the tissue border is the border of a surface of the heart.
19. The method of claim 17 , wherein the border is the endocardium of the heart.
20. The method of claim 1 , wherein the ultrasonic image comprises an image of a chamber of the heart which includes an apex; and wherein the step of locating comprises: defining a first line along a portion of the septal wall of the heart chamber proximate to the apex; defining a second line along a portion of the lateral wall of the heart chamber proximate to the apex; and bisecting the angle between the first and second lines to locate the apex.
21. The method of claim 20 , further comprising a step of analyzing the pixel intensity along the bisected angle to locate the apex of the heart chamber.
22. The method of claim 21 , wherein the heart chamber is the left ventricle.
23. A method of defining a tissue border in a medical ultrasound image, comprising: acquiring a 3-D ultrasound volumetric image; identifying at least three key points or landmarks within the 3-D volumetric image; locating and identifying a center scanline in the ultrasound image relative the three key points or landmarks; determining an array of threshold values correlated to the center scan line as a function of image depth; fitting a trace to a tissue border related to the center scanline and the array of threshold values; and displaying an ultrasonic image with a fitted trace displayed on the tissue border.
24. The method of claim 23 , wherein the step of locating comprises matching a geometric template corresponding to the geometric relationship of the at least the key points or landmarks in the ultrasound image to the scanline.
25. The method of claim 23 , wherein the at least three key points or landmarks are input automatically.
26. The method of claim 23 , wherein the at least three key points or landmarks are input by a user.
27. The method of claim 23 , wherein the step of fitting includes determining a series of chord projections out from the scanline to the myocardium, which chord projections are parallel and occur at regular intervals.
28. The method of claim 27 , further including a step of evaluating each pixel value of a particular chord projection, beginning with the chord pixel value at the center scanline, and if the intensity value is below the threshold for N consecutive pixels, qualifying an area comprising the N values as an edge.
29. The method of claim 28 , wherein at a fixed distance from the origin of each chord projection, small neighborhoods of pixel intensity values are summed and a mean value is determined.
30. The method of claim 29 , wherein a threshold is determined as a percentage of the mean value found for each chord projection.
31. The method of claim 28 , wherein the step of evaluating requires that a contour model of degree N be fit through all of the threshold-determined points of a particular edge, calculating a mean distance between a set of points defining the contour model and actual threshold-determined points and eliminating outlier points which further than a factor F times the mean distance.
32. The method of claim 31 , further including a step for determining the best polynomial approximation to the real edges through the use of a maximization function.
33. The method of claim 32 , wherein the maximization function uses the mean intensity along the contour model, the number of threshold points used to approximate the contour model and the chi-squared error of the polynomial approximation to the threshold points.
34. The method of claim 33 where the chi-squared approximation may be substituted for an arbitrary “goodness-of-fit” estimate.
35. The method of claim 33 where the maximization function can contain any number of parameters.
36. The method of claim 23 , wherein the ultrasound image is a cardiac image and the three key points define the mitral valve and apex of the LV cavity.
37. The method of claim 36 , wherein the border is the endocardium of the heart.
38. The method of claim 36 , wherein the border is the epicardium of the heart.
39. The method of claim 23 , wherein the ultrasonic image comprises an image of a chamber of the heart which includes an apex, and wherein locating comprises: defining a first line along a portion of the septal wall of the heart chamber proximate to the apex; defining a second line along a portion of the lateral wall of the heart chamber proximate to the apex; and bisecting the angle between the first and second lines to locate the apex.
40. The method of claim 39 , further comprising analyzing the pixel intensity along the bisected angle to locate the apex of the heart chamber.