Measurement of vetricle volumes with cardiac MRI

• US 5,273,040 A
• Filed: 11/14/1991
• Issued: 12/28/1993
• Est. Priority Date: 11/14/1991
• Status: Expired due to Term

• Current Assignee: Picker International, Inc. (Marconi Holdings Company Limited)
• Original Assignee: Picker International, Inc. (Marconi Holdings Company Limited)
• Inventors: Wood, Christopher H., NessAiver, Moriel S., Apicella, Anthony
• Primary Examiner(s): Cohen, Lee S.
• Assistant Examiner(s): Pfaffle, K. M.
• Application Number:
  US07/791,855
• Time in Patent Office: 775 Days
• Field of Search: 128/653.1, 128/653.2, 128/695, 382/6, 382/18, 382/22, 364/413.13, 364/413.16, 364/413.18, 358/111
• US Class Current: 600/410
• CPC Class Codes: A61B 5/0263 using NMR
• CPC Class Codes: A61B 5/055 involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
• CPC Class Codes: A61B 5/1076 for measuring dimensions inside body cavities, e.g. using catheters A61B3/1005 takes precedence
• 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 33/5608 Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels image data processing or generation, in general G06T
• CPC Class Codes: G01R 33/5617 using RF refocusing, e.g. RARE
• CPC Class Codes: G01R 33/5619 by temporal sharing of data, e.g. keyhole, block regional interpolation scheme for k-Space [BRISK]
• CPC Class Codes: G01R 33/563 of moving material, e.g. flow contrast angiography
• CPC Class Codes: G06T 2207/20224 Image subtraction
• CPC Class Codes: G06T 7/62 of area, perimeter, diameter or volume
• CPC Class Codes: G06T 7/66 of image moments or centre of gravity

Abstract

A first image (20) and a second image (22) are taken through the patient'"'"'s heart region at small time displaced intervals. The first and second images are subtracted (24) to generate a difference image which is indicative of the tissue which has moved during the short time interval, i.e. the boundary of the ventricles. Voxels from regions outside the boundary are adjusted to remove lung tissue (40), and ventricle boundary or edge voxels (48) and analyzed to generate a non-blood tissue histogram (62). Voxels within the boundary are analyzed to generate a blood tissue histogram (60). The histograms are fit (66, 74) to smooth curves which represent the probability distribution or confidence that each voxel value represents blood or non-blood tissue. Contiguous voxels within the boundary are counted (96) and adjusted for voxel size (98) to create an indication of left and right ventricle volume (100l, 100r). In the preferred embodiment, the ventricle volume is determined by summing the confidence value that each voxel within the boundary represents blood.

Claims

1. A method of generating outlines of ventricular volumes, the method comprising:

non-invasively examining a region of a patient which includes at least one ventricle of a heart, the region being divided into an array of voxels, and generating a first image representation having an intensity value representing each voxel of the region;

storing the first image representation;

a time interval after generating the first image representation, non-invasively re-examining the region and generating a second image representation having an intensity value for each voxel of the region;

storing the second image representation;

subtracting the intensity values corresponding to common voxels of the first and second image representations to create difference intensity values of a difference image representation in which the difference intensity values corresponding to voxels containing substantially stationary tissue are substantially zero and the difference intensity values corresponding to voxels representing tissues which moved in the time interval between the first and second image representations are not zero, whereby the non-zero intensity values define an edge around the at least one ventricle of the heart.

2. A method of generating outlines of ventricular volumes, the method comprising:

non-invasively examining a region of a patient which includes at least one ventricle of a heart and generating (1) a first image representation representing the region and (2) a second image representation of the region using a black blood cine technique such that the first and second image representations depict blood tissue in black and non-blood tissue in lighter shades, the first and second image representations being displaced in time;

subtracting the first and second image representations to create a difference image representation in which stationary tissue is depicted substantially in white and tissues which moved in the time between the first and second image representations are displayed in black, whereby the difference image representation depicts a ventricle defining edge in black against a substantially white background.

3. A method of analyzing ventricular volumes, the method comprising:

non-invasively examining a region of a patient which includes a heart and generating (1) a first image representation having intensity values representing voxels of the region and (2) a second image representation having intensity values representing voxels of the region, the first and second image representations being displaced in time;

subtracting the intensity values of the first and second image representations corresponding to like voxels to create intensity values of a difference image representation in which the intensity values of voxels of the first and second image representations representing substantially stationary tissue cancel and the intensity values of the first and second image representations representing tissues which moved between the first and second image representations do not cancel, such that the non-canceling intensity values define an edge around at least one ventricle of the heart; and

counting a number of voxels within the ventricle defining edge, the number of counted voxels being proportional to ventricular volume.

4. The method as set forth in claim 3 further including:

determining a confidence value indicative of a probability that each intensity value within the ventricle defining edge represents blood; and

weighting each counted voxel in accordance with the confidence value.

5. The method as set forth in claim 4 wherein the step of determining the confidence value includes:

determining a frequency with which each intensity value occurs in voxels of the first image representation within the ventricle defining edge; and

determining a deviation between a most commonly occurring intensity value and each other occurring intensity value, the confidence value being proportional to the deviation.

6. A method as set forth in claim 3 wherein the heart includes a left ventricle and a right ventricle and wherein the counting step includes:

independently counting contiguous voxels of the left and right ventricles within a common generally horizontal plane, whereby in planes in which voxels within the left and right ventricle are contiguous, counting the voxels attributable to both the left and the right ventricle as representing the volume of a single one of the left and right ventricles;

comparing a number of voxels counted in each plane with a number of voxels counted in an adjacent plane to determine whether voxels in both the left and right ventricles have been counted as voxels of the single one of the ventricles.

7. A method as set forth in claim 6 further including:

analyzing the first image representation with an edge operator routine to determine an interface between the left and right ventricles; and

limiting the counting step to voxels on each side of the determined interface.

8. A method of analyzing ventricular volume comprising:

non-invasively examining a region of a patient which includes at least one ventricle to generate values of voxels of a ventricle image representation;

determining boundary indicative voxels of the ventricle image representation which are indicative of a peripheral boundary of the ventricle;

determining a frequency of occurrence for voxel values of inside voxels disposed within the boundary indicative voxels;

in accordance with the frequency of occurrence of each inside voxel value, determining a probability that each inside voxel value corresponds to blood tissue;

counting the inside voxels and weighting each counted inside voxel in accordance with its determined probability to determine a sum which sum is proportional to the ventricular volume.

9. The method as set forth in claim 8 further including:

comparing voxel values of each outside voxel disposed outside the boundary indicative voxels with a voxel value indicative of lung tissue;

eliminating outside voxels whose voxel values indicate lung tissue;

for at least a portion of remaining outside voxels, determining a frequency of occurrence of each outside voxel value;

determining a probability that each remaining outside value corresponds to non-lung, non-blood tissue in accordance with the determined frequency of occurrence of each outside voxel value.

10. The method as set forth in claim 9 wherein in the probability determining step, the frequency of occurrence determinations are fit to curves, which curves are generally bell shaped and wherein the probability is determined in accordance with deviation of each voxel value from a most frequently occurring voxel value.

11. The method as set forth in claim 10 wherein the counting step includes:

determining an approximate center of the ventricle;

comparing each voxel value with the probability curve for blood tissue;

counting each voxel whose voxel value is indicated by the probability curve as probably being blood tissue;

repeating the comparing and counting steps for voxels contiguous to each counted pixel voxel.

12. The method as set forth in claim 10 wherein the curve fitting step includes:

fitting the frequency of occurrence determination for voxels within the ventricle boundary with a first curve;

subtracting the first curve from a frequency of occurrence determination for voxels both inside and outside of the ventricle boundary to create a frequency of occurrence difference; and

fitting a second curve to the frequency of occurrence difference.

13. The method as set forth in claim 12 wherein the curve is one of a Gaussian curve, a lognormal curve, a Maxwell curve, an Erlange curve, a Cauchy curve, a normal curve, and a beta curve.

14. The method as set forth claim 8 wherein the boundary indicative voxels define boundaries of left and right ventricles and wherein some voxels within the boundaries of the left and right ventricles are immediately contiguous, whereby in the counting step, voxels within the boundaries of both the left and right ventricles are counted such that the sum would be indicative of a volume of the left and right ventricles and further including:

determining an inter-ventricle boundary between the left and right ventricles; and

during the counting step, restraining counting to opposite sides of the inter-ventricle boundary such that the left and right ventricle volumes are determined individually.

15. The method as set forth in claim 8 wherein the step of determining the boundary indicative voxels includes:

generating first and second images at a time displaced interval;

subtracting the first and second images to create a difference image representation depicting the ventricle boundary and any other subregions which moved in the interval between the first and second images.

16. The method as set forth in claim 8 further including after determining the boundary indicative voxels, determining a subregion of voxels which encompasses the boundary indicative voxels and non-invasively examining the subregion to generate an image representation of the subregion, whereby resolution within the subregion is improved.

17. The method as set forth in claim 8 wherein the step of determining boundary indicative voxels includes:

generating (1) a first image representation representing the region and (2) a second image representation of the region, the first and second image representations being displaced in time;

subtracting the first and second image representations to create a difference image representation in which voxel values corresponding to substantially stationary tissue cancel and voxel values corresponding to tissues which moved between the first and second image representations do not cancel, whereby the non-canceling voxel values define the peripheral boundary of the ventricle.

18. An apparatus for generating outlines of ventricular volumes, the apparatus comprising:

a non-invasive examination means for non-invasively examining a region of a patient which includes ventricles of a heart, the region being divided into a rectangular array of voxels, and generating (1) a first image representation having an intensity value representing each voxel of the region and (2) a second image representation having an intensity value representing each voxel of the region, the first and second image representations being displaced in time;

a subtracting means for subtracting the intensity values corresponding to common voxels of the first and second image representations to create difference intensity values of a difference image representation in which the difference intensity values corresponding to voxels containing substantially stationary tissue cancel and the difference intensity values corresponding to voxels that contain tissues which moved between the first and second image representations do not cancel, whereby the non-cancelling intensity values of the difference image representation represent an edge around ventricles of the heart; and

a memory means for storing the difference intensity values of the difference image representation.

19. An apparatus for analyzing ventricular volumes, the apparatus comprising:

a non-invasive examination means for non-invasively examining a region of a patient which includes a heart and generating (1) a first image representation representing the region and (2) a second image representation of the region, the first and second image representations being displaced in time;

a subtracting means for subtracting the first and second image representations to create a difference image representation in which voxel values of voxels of the first and second image representations representing substantially stationary tissue cancel and voxel values of like voxels representing tissues which moved between the first and second image representations do not cancel, whereby the difference image representation represents a surface of non-canceling voxel values around ventricles of the heart;

a counting means for counting a number of voxels within the surface of non-canceling voxel values, the number of counted voxels being proportional to ventricular volume; and

,

a display means for displaying the ventricular volume.

20. An apparatus for analyzing ventricular volumes, the apparatus comprising:

a non-invasive examination means for non-invasively examining a region of a patient which includes a heart and generating (1) a first image representation representing the region and (2) a second image representation of the region, the first and second image representations being displaced in time;

a subtracting means for subtracting the first and second image representations to create a difference image representation in which voxel values of voxels of the first and second image representations representing substantially stationary tissue cancel and voxel values of like voxels representing tissues which moved between the first and second image representations do not cancel, whereby the difference image representation represents a ventricular edge.

a probability determining means for determining a confidence value indicative of a probability that each voxel value within the ventricular edge defines blood; and

a weighting means for weighting each counted voxel in accordance with the confidence value.

21. An apparatus for analyzing ventricular volumes comprising:

a non-invasive examination means for non-invasively examining a region of a patient which includes at least one ventricle to generate intensity values of voxels of a ventricle image representation;

a means for determining which voxels of the ventricle image representation are indicative of a peripheral boundary of the ventricle;

a histogram generating means for determining a frequency of occurrence of the intensity values of voxels surrounded by the ventricle boundary;

a means for determining from the frequency of occurrence a probability that each of the intensity values corresponds to blood tissue;

a counting means for counting voxels surrounded by the ventricle boundary and weighting each counted voxel in accordance with its determined probability to determine a sum which is proportional to the ventricular volume.

22. The apparatus as set forth in claim 21 further including:

a means for comparing intensity values of voxels outside of the ventricle boundary with an intensity value indicative of lung tissue;

a means for subtracting outside voxels whose intensity values indicate lung tissue;

a means for determining a frequency of occurrence of each outside voxel intensity value;

a means for determining a probability that each remaining outside voxel intensity value corresponds to non-lung, non-blood tissue in accordance with the determined frequency of occurrence.

23. The apparatus as set forth in claim 22 wherein the means for determining the ventricle boundary includes:

a means for subtracting first and second images generated at a time displaced interval to create a difference image representation depicting the ventricle boundary.

24. An apparatus for measuring ventricular volumes, the apparatus comprising:

a non-invasive examination means for generating volume image representations of a patient region including its heart;

a means for locating a subregion encompassing the heart;

a means for determining a range of intensity values of voxels of the volume image representations, one of the volume image representations having a range of intensity values corresponding to blood tissue and a range of intensity values corresponding to non-blood tissue;

a means for counting contiguous voxels with intensity values in the intensity value range corresponding to blood tissue, which voxel count is indicative of ventricular volume.

25. The apparatus as set forth in claim 24 wherein the locating means includes a subtracting means for subtracting two of the volume image representations which are generated at a time displaced interval.

26. The apparatus as set forth in claim 24 wherein the range determining means includes a means for fitting a distribution of intensity values in each range to blood and non-blood tissue distribution curves.

27. The apparatus as set forth in claim 24 wherein:

the non-invasive examination means generates intensity values of the voxels of the volume image representations;

the subregion locating means includes a means for determining which voxels of the volume image representations are indicative of a peripheral boundary of the ventricle;

the range of values determining means includes a histogram generating means for determining a frequency of occurrence of intensity values of voxels surrounded by the ventricle boundary and a means for determining from the frequency of occurrence a probability that each intensity value corresponds to blood tissue; and

the counting means counts voxels surrounded by the ventricle boundary and wights each counted voxel in accordance with its determined probability to determine a sum which is proportional to the ventricular volume.

28. The apparatus as set forth in claim 27 wherein:

the non-invasive examination means generates (1) a first image representation representing the region and (2) a second image representation of the region, the first and second image representations being displaced in time; and

the ventricle boundary voxel determining means includes a subtracting means for subtracting the first and second image representations to create a difference image representation such that intensity values of voxels of the first and second image representation representing substantially stationary tissue cancel and intensity values of voxels representing tissue which moved between the first and second image representations do not cancel, whereby the difference image representation represents the peripheral boundary of the ventricle.