Normal and abnormal tissue identification system and method for medical images such as digital mammograms

• US 6,310,967 B1
• Filed: 04/29/1998
• Issued: 10/30/2001
• Est. Priority Date: 04/29/1998
• Status: Expired due to Term

• Current Assignee: SETI Institute, University of South Florida (State University System of Florida)
• Original Assignee: University of South Florida (State University System of Florida)
• Inventors: Heine, John J., Cullers, David Kent, Clarke, Laurence P., Stauduhar, Richard Paul, Deans, Stanley R.
• Primary Examiner(s): Johns, Andrew W.
• Assistant Examiner(s): NAKHJAVAN, SHERVIN K
• Application Number:
  US09/069,473
• Time in Patent Office: 1,280 Days
• Field of Search: 382/128, 382/132, 382/270, 382/298, 382/299, 382/302, 382/199, 378/37, 378/62, 128/922, 358/261.3, 706/20, 250/369
• US Class Current: 382/128
• CPC Class Codes: G06T 7/0012 Biomedical image inspection

Abstract

A system and method for analyzing a medical image to determine whether an abnormality is present, for example, in digital mammograms, includes the application of a wavelet expansion to a raw image to obtain subspace images of varying resolution. At least one subspace image is selected that has a resolution commensurate with a desired predetermined detection resolution range. A functional form of a probability distribution function is determined for each selected subspace image, and an optimal statistical normal image region test is determined for each selected subspace image. A threshold level for the probability distribution function is established from the optimal statistical normal image region test for each selected subspace image. A region size comprising at least one sector is defined, and an output image is created that includes a combination of all regions for each selected subspace image. Each region has a first value when the region intensity level is above the threshold and a second value when the region intensity level is below the threshold. This permits the localization of a potential abnormality within the image.

Claims

1. A method for analyzing a medical image to determine whether the image is classifiable as normal, the method comprising the steps of:

applying a wavelet expansion to a digital representation of a raw image, the raw image comprising an array of sectors, each sector having an intensity level, to obtain a plurality of expansion images of varying resolution;

selecting at least one expansion image having a resolution commensurate with a desired predetermined detection resolution range;

dividing each expansion image into a plurality of regions, each region comprising at least one sector; and

creating an output image comprising a combination of all regions for each selected expansion image, each region having a first value when the region intensity level is above a predetermined threshold level and a second value when the region intensity level is below the threshold level, for localizing a potential abnormality within the image;

wherein an absence of a predetermined number of regions having a first value intensity level is indicative of the image being classifiable as normal.

2. The method recited in claim 1, further comprising the steps, prior to the output image creating step, of:

applying a wavelet expansion to a digital representation of a raw control image comprising an array of sectors, each sector having an intensity level, to obtain a plurality of expansion images of varying resolution;

selecting at least one expansion image having a resolution commensurate with a desired predetermined detection resolution range;

determining a functional form of a probability distribution function for each selected expansion image;

determining an optimal statistical normal image region test for each selected expansion image;

establishing the threshold level for the probability distribution function from the optimal statistical normal image region test for each selected expansion image.

3. The method recited in claim 2, wherein the expansion-image selecting step comprises empirically selecting two adjacent expansion images.

4. The method recited in claim 3, wherein the test-determining step comprises forming a search window having a predetermined resolution size for each selected expansion image, and wherein the output image creating step comprises combining the selected two adjacent expansion images and applying a binary mask having the predetermined resolution size to the combined expansion space images.

5. The method recited in claim 2, wherein the optimal statistical normal image region test determining step comprises using a maximum likelihood technique.

6. The method recited in claim 2, wherein the threshold level establishing step comprises applying an iterative procedure to determine a set of operating parameters for each selected expansion image.

7. The method recited in claim 2, wherein the probability density function comprises one of a family of distributions having the form:

$P<br><br>\left(\mathrm{\omega <br><br>}\right)=\frac{k^{2<br><br>N}}{{\left(k^2+{\mathrm{\omega <br><br>}}^2\right)}^N}.$

8. The method recited in claim 2, wherein the test determining step comprises assuming a parametric form of a test statistic distribution.

9. The method recited in claim 1, wherein the test determining step comprises assuming a nonparametric form of a test statistic distribution and using a kernel density estimation method.

10. The method recited in claim 1, wherein the raw image comprises a film image, and further comprising the step, prior to the wavelet-expansion application step, of digitizing the film image.

11. The method recited in claim 1, wherein the wavelet-expansion application step comprises performing a separable kernel two-dimensional pyramid downsampling/upsampling scheme.

12. The method recited in claim 11, wherein each expansion image comprises a difference in information between a first image having a first resolution and a second image having a resolution one-half that of the first image.

13. The method recited in claim 1, wherein the expansion-image selecting step comprises empirically selecting a number of expansion images commensurate with an initial raw image resolution.

14. The method recited in claim 1, wherein the expansion image selecting step comprises selecting a plurality of expansion images, each expansion image having a resolution commensurate with a size range of an abnormality desired to be detected.

15. The method recited in claim 1, further comprising the step of defining a tissue boundary for analysis.

16. The method recited in claim 15, wherein the medical image comprises a digitized mammogram, and wherein the tissue boundary defining step comprises excising a breast boundary on the digitized mammogram.

17. The method recited in claim 16, wherein the breast boundary excising step comprises the steps of:

defining an off-tissue region comprising a generally random noise field and a plurality of anomalous regions;

defining a tissue region comprising signal information;

separating out the random noise field;

determining a remaining contiguous region containing information; and

setting the image region outside the remaining contiguous region to zero.

18. The method recited in claim 1, further comprising the steps of:

defining a cluster area having a size greater than a size of the region;

setting a number of regions within each cluster area in the image having the first value to a first variable;

comparing the first variable with a predetermined second variable;

if the first variable is greater than or equal to the second variable, flagging the cluster area as potentially suspicious for the presence of an abnormality; and

if the first variable is less than the second variable, classifying the cluster area to be normal.

19. The method recited in claim 1, wherein the medical image comprises a digitized mammogram, and the expansion image selecting step comprises selecting two adjacent expansion images having resolution less than 0.5 mm.

20. The method recited in claim 19, wherein the two selected adjacent expansion images further have resolution greater than 0.1 mm.

21. An apparatus for analyzing a medical image to determine whether an abnormality is present comprising:

means for applying a wavelet expansion to a digital representation of a raw image, the raw image comprising an array of sectors, each sector having an intensity level, to obtain a plurality of expansion images of varying resolution;

means for selecting at least one expansion image having a resolution commensurate with a desired predetermined detection resolution range;

means for dividing each expansion image into a plurality of regions, each region comprising at least one sector; and

means for creating an output image comprising a combination of all regions for each selected expansion image, each region having a first value when the region intensity level is above a predetermined threshold level and a second value when the region intensity level is below the threshold level, for localizing a potential abnormality within the image.

22. The apparatus recited in claim 21, further comprising:

means for applying a wavelet expansion to a digital representation of a raw control image, the raw control image comprising an array of sectors, each sector having an intensity level, to obtain a plurality of expansion images of varying resolution;

means for selecting at least one expansion image having a resolution commensurate with a desired predetermined detection resolution range;

means for determining a functional form of a probability distribution function for each selected expansion image;

means for determining an optimal statistical normal image region test for each selected expansion image; and

means for establishing the threshold level for the probability distribution function from the optimal statistical normal image region test for each selected expansion image.

23. The apparatus recited in claim 21, wherein the raw image comprises a film image, and further comprising means for digitizing the raw image.

24. The apparatus recited in claim 21, wherein the wavelet expansion applying means comprises software means.

25. The apparatus recited in claim 21, wherein the output image creating means comprises software means resident within a processor and a screen in electronic communication with the processor for visualizing the output image.