Normal and abnormal tissue identification system and method for medical images such as digital mammograms
First Claim
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 Assignments
0 Petitions
Accused Products
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.
89 Citations
25 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. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20)
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:
-
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. - View Dependent Claims (22, 23, 24, 25)
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.
Specification