Methods and systems for determining breast density

• US 9,895,121 B2
• Filed: 08/19/2014
• Issued: 02/20/2018
• Est. Priority Date: 08/20/2013
• Status: Active Grant

• Current Assignee: Densitas Incorporated
• Original Assignee: Densitas Incorporated
• Inventors: Abdolell, Mohamed, Hope, Tyna, Zaboli, Shiva, Tsuruda, Kaitlyn
• Primary Examiner(s): Tabatabai, Abolfazl
• Application Number:
  US14/912,965
• Publication Number:
  US 20160203600A1
• Time in Patent Office: 1,281 Days
• Field of Search: 382128, 382129, 382130, 382131, 382132, 382133, 382134, 378 37, 378 4, 378 8, 378 21- 27, 378901, 128915, 604 74, 604346, 623 7
• CPC Class Codes: A61B 6/502 for diagnosis of breast, i.e. mammography
• CPC Class Codes: A61B 6/5211 involving processing of medical diagnostic data
• CPC Class Codes: A61B 6/5252 removing objects from field of view, e.g. removing patient table from a CT image
• CPC Class Codes: A61B 6/5258 involving detection or reduction of artifacts or noise
• CPC Class Codes: A61B 6/5294 involving using additional data, e.g. patient information, image labeling, acquisition parameters
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/20061 Hough transform
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30068 Mammography; Breast
• CPC Class Codes: G06T 5/70 Denoising; Smoothing
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

Various embodiments are described herein for methods, devices and systems that can be used to determine a breast density value from an input image of patient'"'"'s breast. In one example embodiment, the breast density value is calculated by receiving an input image corresponding to the digital mammogram, removing metadata information from the input image to generate an intermediate image, generating a region of interest (ROI) image based on the intermediate image, extracting values for predictor variables based on the metadata information, the intermediate image and the ROI image, and calculating breast density based on the values of the predictor variables. The breast density value is calculated based on a breast density model.

Claims

1. A method of determining a breast density measurement from a digital mammogram, the method comprising:

receiving an input image corresponding to the digital mammogram;

removing metadata information from the input image to generate an intermediate image;

generating a region of interest (ROI) image based on the intermediate image;

extracting values for predictor variables from the metadata information, the intermediate image and the ROI image; and

calculating the breast density measurement based on the values of the predictor variables.

2. The method of claim 1, further comprising processing the intermediate image to generate a pre-ROI image, wherein the ROI image is generated by processing the pre-ROI image.

3. The method of claim 2, wherein generating the pre-ROI image comprises at least one of resizing the intermediate image and removing noise.

4. The method of claim 3, wherein the noise is removed by applying smoothing or by removing salt and pepper noise from the intermediate image.

5. The method of claim 2, wherein generating the ROI image comprises generating an initial breast signal mask from the pre-ROI image.

6. The method of claim 5, wherein generating the ROI image comprises refining the initial breast signal mask by removing an exterior part of a nipple area from the pre-ROI image.

7. The method of claim 5, wherein generating the ROI image comprises refining the initial breast signal mask by removing the pectoral muscles which comprises using a Hough transformation.

8. The method of claim 5, wherein generating the ROI image comprises refining the region under a main breast in the intermediate image.

9. The method of claim 8, wherein when the input image is a MLO view of the digital mammogram refining the region under the main breast comprises removing a chest wall under the main breast and when the input image is a CC view of the digital mammogram refining the region under the main breast comprises removing adjacent breast.

10. The method of claim 5, wherein generating the ROI image comprises refining the initial breast signal mask by removing subcutaneous fat from the breast.

11. The method of claim 5, wherein generating the ROI image further comprises refining the initial breast signal mask to remove skin fold regions by removing an arc region containing a skin fold near an inframammary region or by removing an arc region containing a skin fold at the armpit.

12. The method of claim 1, wherein removing metadata information from the input image comprises removing metadata information from the header of the input image.

13. The method of claim 1, wherein the input image is based on a DICOM standard, and removing the metadata information from the input image comprises removing DICOM header information.

14. The method of claim 13, wherein the predictor variables from the metadata information comprise at least one of patient specific parameters, image acquisition device parameters and image acquisition parameters.

15. The method of claim 14, wherein the image acquisition parameters comprises at least one of relative x-ray exposure, exposure duration, laterality, patient orientation and peak kilo-voltage, the patient specific parameters comprise at least one of age of the patient, gender of the patient, patient date of birth, and patient identification information, and the image acquisition device parameters comprise at least one of name and model of the device used to obtain the digital mammogram.

16. The method of claim 1, wherein measurements for predictor variables are generated from a region of the intermediate image corresponding to a sub-set of non-zero pixel values of the ROI image.

17. The method of claim 1, wherein the predictor variables comprise histogram features including at least one of skewness, entropy, maximum between class variance, ninety nine percent index and fifty percent index.

18. The method of claim 1, wherein predictor variables comprise texture features including at least one of average gradient magnitude value of the intermediate image within the region defined by the non-zero values of the ROI image, average gradient magnitude value of the intermediate image within a subset of the region defined by the non-zero values of the ROI image, edge ratio of the intermediate image in the region defined by the non-zero values of ROI image, and edge ratio of the intermediate image within a subset of the region defined by the non-zero values of ROI image.

19. The method of claim 1, wherein extracting values for predictor variables from the input image, the intermediate image and the ROI image comprises extracting values for predictor variables based on at least one of relative x-ray exposure, exposure duration, peak kilo-voltage, skewness, entropy, maximum between class variance, ninety nine percent index and fifty percent index, average gradient magnitude value of the intermediate image within the region defined by the non-zero values of the ROI image, average gradient magnitude value of the intermediate image within a subset of the region defined by the non-zero values of the ROI image, edge ratio of the intermediate image in the region defined by the non-zero values of ROI image, and edge ratio of the intermediate image within a subset of the region defined by the non-zero values of ROI image.

20. The method of claim 1, wherein the breast density measurement is determined using a breast density model developed based on a regression analysis method, a statistical learning method, or a machine learning method.

21. The method of claim 20, wherein the breast density model is based on one of regression analysis, linear regression, multiple adaptive regression splines, classification and regression trees, random forests, artificial neural networks, support vector machines and ensemble learning algorithms.

22. The method of claim 1, wherein the input image comprises a processed raw digital mammogram image or a ‘

for presentation’

digital mammogram image.

23. A system for determining a breast density value from a mammogram image of a breast, the system comprising:

a memory unit; and

a processing unit coupled to the memory unit, the processor unit being configured to;

receive an input image corresponding to the digital mammogram;

remove metadata information from the input image to generate an intermediate image;

generate a region of interest (ROI) image based on the intermediate image;

extract values for predictor variables from the metadata information, the intermediate image and the ROI image; and

calculate the breast density measurement based on the values of the predictor variables.

24. The system of claim 23, wherein the processing unit is configured to process the intermediate image to generate a pre-ROI image and to generate the ROI image by generating an initial breast signal mask from the pre-ROI image.

25. The system of claim 24, wherein the processing unit is configured to generate the ROI image by performing at least one of:

removing an exterior part of a nipple area from the pre-ROI image, removing the pectoral muscles, refining a region under a main breast in the intermediate image, by removing subcutaneous fat from the breast, and by removing skin fold regions.

26. The system of claim 23, wherein the input image is based on a DICOM standard, and removing the metadata information from the input image comprises removing DICOM header information.

27. The system of claim 26, wherein the predictor variables from the metadata information comprise at least one of patient specific parameters, image acquisition device parameters and image acquisition parameters.

28. The system of claim 27, wherein the image acquisition parameters comprises at least one of relative x-ray exposure, exposure duration, laterality, patient orientation and peak kilo-voltage, the patient specific parameters comprise at least one of age of the patient, gender of the patient, patient date of birth, and patient identification information, and the image acquisition device parameters comprise at least one of name and model of the device used to obtain the digital mammogram.

29. The system of claim 23, wherein measurements for predictor variables are generated from a region of the intermediate image corresponding to a sub-set of non-zero pixel values of the ROI image, and the predictor variables comprise at least one of histogram features and texture features.

30. The system of claim 23, wherein the breast density measurement is determined using a breast density model developed based on a regression analysis method, a statistical learning method, or a machine learning method.

31. A non-transitory computer-readable storage medium storing computer-executable instructions, the instructions for causing a processor to perform a method of determining a breast density value from a mammogram image of a breast, the method comprising:

receiving an input image corresponding to the digital mammogram;

removing metadata information from the input image to generate an intermediate image;

generating a region of interest (ROI) image based on the intermediate image;

extracting values for predictor variables from the metadata information, the intermediate image and the ROI image; and

calculating the breast density measurement based on the values of the predictor variables.