Quantitative analysis system and method for certifying ultrasound medical imaging equipment

• US 6,370,480 B1
• Filed: 05/21/1999
• Issued: 04/09/2002
• Est. Priority Date: 05/21/1999
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Gupta, Rajiv, Dailey, Christopher James, Zheng, Yibin
• Primary Examiner(s): Hoff, Marc S.
• Assistant Examiner(s): Raymond, Edward
• Application Number:
  US09/316,011
• Time in Patent Office: 1,054 Days
• Field of Search: 702/39, 702/54, 702/159, 702/171, 702/71, 702/33, 702/35, 73/10.1, 73/18.2, 738/659, 736/18, 736/19, 736/40, 735/70, 736/20, 738/664, 738/656, 128/915, 128/916, 128/922, 128/923
• US Class Current: 702/39
• CPC Class Codes: A61B 8/00 Diagnosis using ultrasonic, sonic or infrasonic waves imaging of objects using sonar G01S15/00
• CPC Class Codes: A61B 8/587 Calibration phantoms models for medical purposes G09B23/28
• CPC Class Codes: G01S 7/5205 Means for monitoring or calibrating
• CPC Class Codes: G06T 2207/30168 Image quality inspection
• CPC Class Codes: G06T 7/0002 Inspection of images, e.g. flaw detection

Abstract

An analysis system and method provide for quantitatively evaluating image quality characteristics of an ultrasound imaging machine that evaluates at least one image representation of a standard phantom acquired by the image machine. The machine under test by comparing acquired parameters with prestored values, and returning a determined set of image quality indices, along with a single index representing an arithmetic combination of all other image quality indices, which indicate the accuracy of the test image relative to a “gold standard” that has been pre-established for the model of imaging machine under investigation. The system, which includes a computer-programmed set of instructions and data, optionally includes at least one standard phantom. The image quality indices, or metrics, quantitatively represent an evaluation of a test image using a set of relatively subjective criteria that include homogeneity, contrast, signal attenuation and penetration of depth, pin to background ratio in near and far-field, axial and lateral resolution, modulation transfer function, and geometric distortion, and axial and lateral linearity. These image quality indices are determined by specific algorithms and then combined to form an image health index. The image health index and the individual component indices are compared to a gold standard set of indices obtained from an equivalent imaging machine operating under optimum conditions and settings.

Claims

1. A method of quantitatively evaluating image quality characteristics of an ultrasound imaging machine comprising the steps of:

storing an optimum ultrasound image representation of a phantom;

storing a test ultrasound image representation of said phantom, said test ultrasound image representation being obtained from an ultrasound instrument under test;

electronically and automatically shifting in memory the stored test ultrasound image representation until said representation is in registration with said optimum ultrasound image representation

comparing one or more image quality parameters of the test ultrasound image with one or more regions of interest of the optimum stored ultrasound image representation with one or more corresponding regions of the test ultrasound image representation, each said one or more regions of interest corresponding to an image quality parameter to be evaluated;

calculating from the comparison a respective image quality index for each of said one or more regions of interest; and

utilizing said image quality indices to calculate an image health index from said respective image quality indices, wherein said image health index quantitatively represents image quality characteristics of the ultrasound imaging machine under test.

2. The method of claim 1, wherein said shifting the test ultrasound image representation includes calculation of a cross-correlation coefficient, said cross-correlation coefficient representing a quantitative degree of proximity of coordinates of features in said test image representation relative to coordinates of corresponding features in said optimum image representation.

3. The method of claim 2, wherein said cross-correlation index is given by $C=\frac{\sum _{}^{}}{}$

i,j 



I1

(i,j)

 



I2

(i,j)∑

i,j

I12

(i,j)

∑

i,j

 



I22

(i,j)

wherein C equals the cross-correlation coefficient, I₁(i,j) equals test image pixel locations, and I₂(i,j) equals optimum image pixel locations.

4. The method of claim 1, wherein said image quality index is a homogeneity image quality index calculated by:

selecting a region of the image of said test ultrasound

image representation that contains a uniform background area with no target present, the region having a total area;

repeatedly dividing said total area into a selected number of blocks, each block having a pixel brightness;

calculating an average pixel brightness for each said block, wherein the averages constitute a multi-resolution scale of brightness;

calculating a ratio of the mean and standard deviation of said the average pixel brightness for each said block, wherein said ratio represents homogeneity of said each block; and

calculating a weighted average of all said ratios, wherein said weighted average is said homogeneity index.

5. The method of claim 4, wherein said step of repeatedly dividing said total area into a selected number of blocks comprises repeatedly dividing said total area into blocks until a block size of one pixel is achieved.

6. The method of claim 1, wherein said image quality index is a contrast image quality index calculated by:

selecting an exact region of interest including grayscale targets;

computing a mean brightness value for each grayscale target;

selecting for each grayscale target a stored respective nominal brightness value for corresponding grayscale target associated with said optimum ultrasound image representation; and

computing a total squared error of a difference between all said mean brightness values and said corresponding stored nominal values, said total squared error being the contrast image quality index.

7. The method of claim 6, wherein said step of selecting an exact region of interest comprises:

identifying center coordinates and respective radii of all grayscale targets in said optimum image representation;

translating said center coordinates and respective radii to said test image representation;

selecting an approximate region of interest of said test ultrasound image representation; and

selecting at least one grayscale target center coordinate within said approximate region of interest and generating a mask based on all said selected grayscale target center coordinates and respective radii, said mask being an exact region of interest.

8. The method of claim 6, wherein said step of computing a total squared error comprises:

computing a mean brightness value for each said selected grayscale target within said exact region of interest;

computing a difference between each said mean brightness value and a respective corresponding stored nominal brightness value, wherein said respective corresponding stored nominal brightness values are obtained from said optimum image representation; and

computing a total squared error of a difference between all said mean brightness values and said respective corresponding stored nominal values, said total squared error being the contrast image quality index.

9. The method of claim 1, wherein said image quality index is a signal attenuation image quality index and penetration depth image quality index calculated by:

selecting a region of the image of said test ultrasound image representation that contains a uniform background area;

calculating an average value of pixel values for at least one depth x in said region; and

calculating an estimate of a signal attenuation image quality index λ

based on said average value, wherein 1/λ

is the penetration depth index.

10. The method of claim 9, wherein λ

is a signal attenuation index, I₀ an average of pixel values for a depth x, and an intensity profile I(x) is a mean pixel brightness function of x for every depth x according to the intensity profile I(x)=I₀e<sup>−

λ

X</sup>, said calculating an estimate of a signal attenuation index λ

comprises;

calculating an average of pixel values I₀ for each depth x;

performing a sequence of linear regression steps to calculate an estimate of the signal attenuation index λ

, said steps including

calculating I(x)=I₀e−

<sup>−

λ

X</sup>;

calculating log I(x); and

forming a plot of log I(x) versus x and drawing a straight line having a slope through said plot of log I(x) versus x, wherein a negative value of said slope is an estimate of the signal attenuation image quality index λ and

1/λ

is the penetration depth image quality index.

11. The method of claim 1, wherein said image quality index is a pin to background ratio image quality index calculated by:

selecting an exact region of the image of said test ultrasound image representation that includes evenly spaced, vertical pins imbedded at specific depths in a background gel;

calculating a first average brightest pixel value in said selected region, said first average brightest pixel value representing a pin;

calculating a second average pixel brightness value of a pixel region contiguous to said pin; and

calculating a ratio of said first average brightest pixel value to said second average pixel brightness value, wherein the ratio is the pin to background ratio image quality index.

12. The method of claim 11, wherein said step of selecting an exact region of interest comprises:

identifying center coordinates and respective radii of all pin targets in said optimum image representation;

translating said center coordinates and respective radii to said test image representation;

selecting an approximate region of interest of said test ultrasound image representation; and

selecting at least one pin target center coordinate within said approximate region of interest and generating a mask based on said at least one pin target center coordinate and respective radii, said mask being an exact region of interest.

13. The method of claim 1, wherein said image quality index is a resolution image quality index calculated by:

selecting a region of interest within said test ultrasound image representation that includes at least two evenly spaced pins arranged along a vertical axis and at least two evenly spaced pins arranged along a horizontal axis;

selecting brightest pixels within said region of interest, each said brightest pixel representing a respective center of a pin;

calculating a first half dimension and a first maximum dimension between adjacent pin centers arranged along a vertical axis and a second half dimension and a second maximum dimension between adjacent pin centers arranged along a horizontal axis; and

calculating a first ratio of said first half width dimension to said first maximum dimension for each said adjacent pin centers arranged along a vertical axis and calculating a second ratio of said second half width dimension to said second maximum dimension for each said adjacent pin centers arranged along a horizontal axis, wherein said first ratio is a vertical component of the resolution index and said second ratio is a horizontal component of the resolution index.

14. The method of claim 1, wherein said image quality index is a geometric distortion image quality index calculated by:

selecting at least one uniform grayscale target within a region of the ultrasound test image representation;

determining a vertical and a horizontal radius for said at least one uniform grayscale target;

calculating a ratio of each said vertical radius to a respective horizontal radius; and

calculating a total squared error of said ratios, said error being said geometric distortion image quality index.

15. The method of claim 14, wherein said step of selecting at least one uniform grayscale target comprises:

identifying center coordinates of all first grayscale targets in said optimum image representation;

translating all said center coordinates to corresponding second grayscale targets in said test ultrasound image representation;

selecting an approximate region of interest of said test ultrasound image representation, wherein said selected approximate region includes at least one second grayscale target, each at least one second grayscale target having a translated center coordinate; and

selecting at least one second grayscale target center coordinate within said approximate region of interest.

16. The method of claim 15, wherein said step of determining a vertical and a horizontal radius comprises:

detecting changes in average pixel values in any direction from each said second grayscale target center coordinate;

selecting for each said second grayscale target a first average pixel value based on said detected changes, wherein said first average pixel value corresponds to a pixel oriented horizontally relative to said second grayscale target center coordinate and said first average pixel value represents an edge of a second grayscale target;

selecting for each said second grayscale target a second average pixel value based on said detected changes, wherein said second average pixel value corresponds to a pixel oriented vertically relative to said second grayscale target center coordinate and said second average pixel value represents an edge of a second grayscale target; and

determining a horizontal radius for each said first average pixel value and determining a vertical radius for each said second average pixel value, each said radius being a distance from a corresponding center coordinate of a respective second grayscale object to a respective selected average pixel.

17. The method of claim 1, wherein said image quality index is a horizontal and a vertical linearity image quality index comprising:

selecting a region of the test ultrasound image representation that includes a representation of a set of evenly spaced pins arranged along a vertical axis and a set of evenly spaced pins arranged along a horizontal axis, each set being imbedded at a selected depth in a background gel;

determining a first set of coordinates of pins of said test ultrasound image representation;

determining a second set of coordinates of corresponding pins of said optimum image representation; and

calculating a first total squared error of differences between said first and second sets of respective coordinates of pins arranged along a vertical axis, and calculating a second total squared error of differences between said first and second sets of respective coordinates of pins arranged along a horizontal axis, wherein said first and second total squared errors are the vertical linearity image quality index and horizontal linearity image quality index, respectively.

18. The method of claim 17, wherein said step of

determining a first set of coordinates of pins of said test ultrasound image representation and said step of determining a second set of coordinates of corresponding pins of said optimum image representation comprises:

selecting first brightest pixel values in said selected region, each said first brightest pixel value representing a center of a respective pin;

determining a first set of coordinates corresponding to each said first brightest pixel values, said first set being determined separately for said pins arranged along a horizontal axis and for said pins arranged along a vertical axis; and

determining a second set of coordinates of second brightest pixel values from said optimum image representation, said second brightest pixel values corresponding to said first brightest pixel values, said second set being determined separately for said pins arranged along a horizontal axis and for said pins arranged along a vertical axis.

19. The method of claim 1, wherein said image health quality index is obtained by arithmetically combining at least two of said image quality indices, wherein said image quality indices include said homogeneity index, said contrast index, said signal attenuation and penetration depth index, said pin to background ratio index, said resolution index, said linearity index, and said geometric distortion index.

20. The method of claim 19, wherein said step of arithmetically combining comprises determining the Health Quality Index by performing the following calculation $\sqrt{\frac{{(x_1-{\mu }_{}}^{}}{}}$

1)2σ

12+…



 

+(xn-μ

n)2σ

n2

wherein x₁. . . x_n are values of respective image quality indices calculated for a test ultrasonic image representation, μ

₁. . . μ

_n are values of respective image quality indices calculated for an optimum image representation, and σ

₁. . . σ

_n are standard deviations of the x₁. . . x_n values.

21. A system for quantitatively evaluating image quality characteristics of an ultrasound imaging machine comprising:

a control circuit including a storage memory for data storage;

a standard phantom;

a system control program included in said control circuit for controlling an optimum ultrasound image representation of said phantom and a test ultrasound image representation of said phantom, said test ultrasound image representation being obtained from an ultrasound instrument under test, wherein one or more image quality parameters to be calculated and one or more regions of interest of the test ultrasound image representation are selected, each of said one or more regions of interest corresponding to an image quality parameter to be tested;

the system control program automatically shifting in memory the test ultrasound image representation until said representation is in registration with said optimum ultrasound image representation, the system control program calculating respective image quality indices for each said selected one or more regions of interest, and further calculating an image health index from said respective image quality indices, said image health index quantitatively representing image quality characteristics of an ultrasound imaging machine.

22. The system of claim 21, wherein said system control program automatically shifts in memory the test ultrasound image representation until said representation is in registration with said optimum ultrasound image representation by calculating a cross-correlation coefficient, said cross-correlation coefficient representing a quantitative degree of proximity of said test ultrasound image representation relative to said optimum image registration and is given by $C=\frac{\sum _{}^{}}{}$

i,j 



I1

(i,j)

 



I2

(i,j)∑

i,j

I12

(i,j)

∑

i,j

 



I22

(i,j)

wherein C equals the cross-correlation coefficient;

I₁(i,j) equals test image pixel locations, and I₂(i,j) equals optimum image pixel locations.

23. The system of claim 22, wherein said system control program calculates said image health quality index by causing at least two of said image quality indices to be arithmetically combined in said control, wherein said image quality indices include a homogeneity index, a contrast index, a signal attenuation and penetration depth index, a pin to background ratio index, a resolution index, a linearity index, and a geometric distortion index.

24. The system of claim 23, wherein said system control program (101) causes said image health quality index to be calculated by said control circuit by performing the following calculation $\sqrt{\frac{{(x_1-{\mu }_{}}^{}}{}}$

1)2σ

12+…



 

+(xn-μ

n)2σ

n2

wherein x₁. . . x_n are values of respective image quality indices calculated for a test ultrasonic image representation, μ

₁. . . μ

_n are values of respective image quality indices calculated for an optimum image representation, and σ

₁. . . σ

_n are standard deviations of the x₁. . . x_n values.