Transillumination method apparatus for the diagnosis of breast tumors and other breast lesions by normalization of an electronic image of the breast

• US 5,079,698 A
• Filed: 05/03/1989
• Issued: 01/07/1992
• Est. Priority Date: 05/03/1989
• Status: Expired due to Fees

• Current Assignee: Advanced Light Imaging Technologies Limited
• Original Assignee: Advanced Light Imaging Technologies Limited
• Inventors: Orth, Paul H., Grenier, Leonard E., Funt, Brian V., McIntosh, Donald M. F.
• Primary Examiner(s): Hayes, Gail O.
• Application Number:
  US07/346,853
• Time in Patent Office: 979 Days
• Field of Search: 364/413.13, 364/413.16, 364/413.17, 364/518, 364/521, 382/6, 382/50, 382/54
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10132 Ultrasound image
• CPC Class Codes: G06T 2207/30068 Mammography; Breast
• CPC Class Codes: G06T 2207/30096 Tumor; Lesion
• CPC Class Codes: G06T 5/20 using local operators
• CPC Class Codes: G06T 5/94 based on local image properties, e.g. for local contrast enhancement
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/12 Edge-based segmentation

Abstract

A method and apparatus for enhancing the contrast of a local area of interest within an electronic image of an object, such as a female breast, which has been transilluminated by non-ionizing radiation such as light or sound. The area of interest may be a cancerous tumor, cyst or another object which differentially absorbs or transmits the radiation. Enhancement of contrast is by normalization of the electronic image. Normalization includes modeling the illumination field of the image to compensate for the non-uniformity of the illumination field, and then combining the modeled field with the original image. Four normalization processes are disclosed: Gaussian curve fitting, geometric mean smoothing, arithmetic mean smoothing, and arithmetic mean smoothing only within the boundary of the local area of interest. Also disclosed is a process for highlighting local areas which are the result of enhanced transmission of the radiation, such as potential cyst sites, and color mapping more than one displayed image. The normalized image may be displayed for analysis.

Claims

1. A method of breast imaging for medical diagnosis of the presence of breast lesions comprising,

passing a non-ionizing radiation through the breast in a range of frequencies which are differentially absorbed or transmitted by lesions within the breast,

detecting the radiation which has passed through the breast,

generating an electronic image of the breast,

converting the image so generated into a digital format of pixel brightness values,

electronically normalizing the digital image by compensating for the non-uniformity of the radiation field component of the image to enhance the contrast of areas representing locally high differential absorption or transmission of the radiation through the breast,

and displaying the normalized image for analysis by a medical practitioner.

2. The method according to claim 1 wherein normalization of the digital image includes:

for each line in the array of pixels within the digital format of the image;

converting the numeric value of each pixel to its logarithm (herein "logpixel"),

fitting a gaussian curve to a set of selected, spaced logpixels in the line to generate a set of curve fitted logpixels,

subtracting the curve fitted logpixels from the input image logpixels,

adding a normalization constant,

converting each of the normalization logpixels to its antilog,

and generating a digital image made up of each line of normalized pixels for display.

3. The method according to claim 1 wherein normalization includes:

determining an average pixel value within a preselected local area in the neighborhood of each pixel in the digital input image, the dimensions of said local area being optimized to accentuate the areas representing locally high differential absorption or transmission of the radiation in the digital image of the breast,

then determining the ratio of each pixel in the input image to the corresponding local average pixel,

and scaling the pixel ratio values to allowable gray scale levels for display.

4. The method according to claim 3 wherein the local area average for each pixel is calculated as a geometric mean.

5. The method according to claim 3 wherein the local area average for each pixel is calculated as an arithmetic mean.

6. The method according to claim 3 wherein the optimized local area is a regular polygon.

7. The method according to claim 6 wherein the optimized local area is a square of 65 pixels on a side.

8. The method of claim 3 including reducing distortion at the boundary between the image of the breast and the remainder of the digital image by normalization only within the breast region of the digital image comprising:

detecting the edge of the breast by sampling the occurrence of points of light to dark gradients in the input image,

determining the coordinates of the edge points,

determining the center of the edge points and the average distance of an edge point from the center,

determining a single edge point which is most likely to be a true boundary of the breast region,

then generating a closed curve by rotating an axis through a set of edge points located within a predetermined distance range from the center, starting with said single edge point, and

then normalizing the image within the region defined by the closed curve.

9. The method of claim 3 including the step of further enhancing areas representing locally high absorption by mapping all pixels whose input image to local average pixel ratio is less than a predetermined value to a dark gray value level.

10. The method of claim 9 including the step of falsely coloring the mapped pixels.

11. The method of claims 2, 3 or 8 including further enhancing the areas representing locally high transmission of the radiation to aide in determining cyst sites, comprising:

taking the average of preselected sets of normalized pixels to smooth the contours that represent the cyst regions,

assigning each pixel having a gray scale value above a threshold value to one of a preselected number of gray values,

then generating a contour region by selecting a set of pixels in a line of pixels, each of which has both a gray value above the threshold value and a value different from the other pixels,

setting each of the selected pixels to a contour line value to generate a contour region, and then overlaying the normalized image with the contour region.

12. The method of claim 11 including the step of coloring the contour region, and leaving the rest of the image uncolored.

14. The method according to claims 1, 2, 3, 8 or 13 wherein the radiation is light having a wavelength in the red to near infrared region of the electromagnetic spectrum.

15. The method according to claim 14 wherein the wavelength of the radiation is 650 nM to 950 nM.

16. The method according to claims 1, 2, 3, 8 or 13 wherein the radiation is ultrasound.

17. The method according to claim 16 wherein the radiation has an ultrasound signal frequency of 7.5 MHz

18. The method according to claims 1, 2, 3, 8 or 13 wherein the radiation is either ultrasound or light in the red to near infrared region of the electromagnetic spectrum.

19. The method according to claim 1, 2, 3, 8 or 13 wherein the normalization is done on a pixel by pixel basis.

20. A method according to claims 1 or 13 wherein normalization includes:

determining an average pixel value within a preselected local area in the neighborhood of each pixel in the digital input image, the dimensions of said local area being optimized to accentuate the areas representing locally high differential absorption or transmission of the electromagnetic radiation in the digital image of the breast,

then determining the ration of each pixel in the input image to the corresponding local average.

21. The method according to claim 20 wherein the local area average for each pixel is calculated as a geometric mean.

22. The method according to claim 20 wherein the local area average for each pixel is calculated as an arithmetic mean.

23. The method of claim 20 including reducing distortion at the boundary between the image of the breast and the remainder of the digital image comprising:

determining the boundary of the breast image, and

then normalizing the image within the region defined by said boundary.

32. The method according to claims 1 or 24 wherein normalization of the image includes:

sampling brightness values of pixels within the digital format,

fitting a Gaussian curve to the set of sampled pixel values to generate a set of curve fitted pixel values,

subtracting the curve fitted pixel values from the pixel values of the digital format to generate a normalized image made up of a pixel format resulting from difference in curve fitted values and pixel brightness values of the digital array.

36. The method according to claim 1, 13 or 24 comprising:

generating two or more normalized images for simultaneous display,

segmenting each image for color mapping of preselected pixel gray values within each image to different mapping schemes,

said segmenting step including adding or subtracting an integer equal to the number of images to be displayed to sets of successive pixel values so that the resulting pixel values within each image are not equal.

13. A method of breast imaging for medical diagnosis of the presence of breast lesions comprising,

passing a non-ionizing radiation through the breast in a range of frequencies which are differentially absorbed or transmitted by lesions within the breast,

detecting the radiation which has passed through the breast,

generating an electronic image of the breast,

converting the image so generated into a digital format of pixel brightness values,

electronically normalizing the digital image by compensating for the non-uniformity of the radiation field component of the image to enhance the contrast of areas representing locally high differential absorption or transmission of the radiation through the breast,

said normalizing step including averaging the gray values representing the electromagnetic radiation field and determining the ratio of the original radiation field to the averaged field,

and displaying the normalized image for analysis by a medical practitioner.

24. A method of enhancing the contrast of a local area of interest within an electronic image resulting from passing radiation through an object which differentially absorbs the radiation, comprising:

converting the image into a digital array of pixel gray values,

normalizing the digital image to compensate for the non-uniformity of the radiation field and to enhance the contrast of areas representing locally high differential absorption of the radiation transmitted through the object,

said normalizing step including modeling the illumination field to approximation of the desired field and then combining the original field with the modeled field,

and displaying the normalized image for analysis.

25. The method of claim 24 wherein normalization of the digital image includes:

for each line in the array of pixels within the digital format of the image;

converting the numeric value of each pixel to its logarithm (herein "logpixel"),

fitting a gaussian curve to a set of selected, spaced logpixels in the line to generate a set of curve fitted logpixels,

subtracting the curve fitted logpixels from the input image logpixels,

adding a normalization constant,

converting each of the normalization logpixels to its antilog,

and generating a digital image made up of each line of normalized pixels for display.

26. A method according to claim 24 wherein normalization includes:

determining an average pixel value within a preselected local area in the neighborhood of each pixel in the digital input image, the dimensions of said local area being optimized to accentuate the areas representing locally high differential absorption or transmission of the radiation in the digital image,

then determining the ratio of each pixel in the input image to the corresponding local average pixel,

and scaling the pixel ratio values to allowable gray scale levels for display.

27. The method according to claim 26 wherein the local area average for each pixel is calculated as a geometric mean.

28. The method according to claim 26 wherein the local area average for each pixel is calculated as an arithmetic mean.

29. The method according to claim 26 wherein the optimized local area is a regular polygon.

30. The method of claim 29 including reducing distortion at the boundary between the image of the object and the remainder of the digital image by normalization only within the object region of the digital image comprising:

detecting the edge of the object region by sampling the occurrence of points of light/dark gradients in the input image,

determining the coordinates of the edge points,

determining the center of the edge points and the average distance of an edge point from the center,

determining a single edge point which is most likely to be a true boundary of the object region,

then generating a closed curve by rotating an axis through a set of edge points located within a predetermined distance range from the center, staring with said single edge point,

and then normalizing the image within the region defined by the closed curve.

31. The method of claims 25, 26 or 30 including further enhancing the areas representing locally high transmission comprising:

taking the average of preselected sets of normalized pixels to smooth the contours that represent the cyst regions,

assigning each pixel having a gray scale value above a threshold value to one of a preselected number of gray values,

then generating a contour region by selecting a set of pixels in a line of pixels, each of which has both a gray value above the threshold value and a value different from the other pixels,

setting each of the selected pixels to a contour line value to generate a contour region, and then overlaying the normalized image with the contour region.

33. The method of claim 26 including reducing distortion at the boundary between the image of the object and the remainder of the digital image comprising:

determining the boundary of the image of the object, and

then normalizing the image of the object within the region defined by said boundary.

34. The method of claim 26 including the step of further enhancing areas representing locally high absorption by mapping all pixels whose input image to local average pixel ratio is less than a predetermined value to a dark gray value level.

35. The method of claim 34 including the step of falsely coloring the mapped pixels.

37. A method according to claim 24 wherein normalization includes:

determining an average pixel value within a preselected local area in the neighborhood of each pixel in the digital input image, the dimensions of said local area being optimized to accentuate the areas representing locally high differential absorption or transmission of the radiation in the digital image,

then determining the ratio of each pixel in the input image to the corresponding local average.

38. Transillumination apparatus for enhancing the contrast of a local area of interest within an electronic image resulting from passing radiation through an object which differentially absorbs the radiation, comprising:

a source of non-ionizing radiation of a frequency which can be transmitted through the object whose image is to be recorded,

a detector for detecting the radiation which has passed through the object and transducing it into an electronic signal representative of the image,

circuit means for converting the signal into a digital format of pixel brightness values,

means for electronically normalizing the digital image by compensating for the non-uniformity of the radiation field component of the image to enhance the contrast of areas representing locally high differential absorption or transmission of the radiation transmitted through the object,

said means for normalizing the digital image including modeling the illumination field to an approximation of the desired field and then combining original light field with the modeled field,

and means for displaying the image.

39. Transillumination apparatus in accordance with claim 38 wherein the means for normalizing the image includes means for sampling the brightness values of pixels within the digital format, fitting a Gaussian curve to the set of sample pixel values to generate a set of curve fitted pixel values, and subtracting the curve fitted pixel values from the pixel values of the digital format to generate a normalized image made up of a pixel format resulting from the difference in the curve fitted values and the pixel brightness values of the digital array.

40. Transillumination apparatus in accordance with claim 38 wherein said normalizing means includes means for determining an average pixel value within a preselected local area in the neighborhood of each pixel in the digital input image, the dimensions of said local area being optimized to accentuate the areas representing locally high differential absorption or transmission of the radiation in the digital image of the object, and means for determining the ratio of the pixel values in the input image to the corresponding local average pixel.

41. Transillumination apparatus in accordance with claim 40 including means for reducing distortion at the boundary between the image of the object and the remainder of the digital image by determining the boundary of the object image and normalizing the image within the region defined by said boundary.

42. Transillumination apparatus in accordance with claims 39 or 40 wherein the radiation is light having a wavelength in the red to near infrared region of the electromagnetic spectrum.

43. Transillumination apparatus in accordance with claims 39 or 40 wherein the radiation is ultrasound.

44. Transillumination apparatus in accordance with claims 39 or 40 including means for enhancing areas representing locally high transmission, said means including means for taking the average of preselected sets of normalized pixels to smooth the contours that represent the areas of locally high transmission, assigning each pixel having a brightness value above a threshold value to one of a preselected number of brightness values, and selecting a set of pixels in a line of pixels, each of which has a brightness value above the threshold value and a value different from the other pixels, and setting each of the selected pixels to a contour line value to generate a contour region, and then overlaying the normalized image with the contour region.