Method and system for detection of suspicious lesions in digital mammograms using a combination of spiculation and density signals

• US 6,198,838 B1
• Filed: 08/27/1998
• Issued: 03/06/2001
• Est. Priority Date: 07/10/1996
• Status: Expired due to Term

• Current Assignee: Hologic Incorporated
• Original Assignee: R2 Technology, Inc. (Hologic Incorporated)
• Inventors: Doi, Takeshi, Te Brake, Guido M., Roehrig, Jimmy R., Karssemeijer, Nico
• Primary Examiner(s): Patel, Jayanti K.
• Application Number:
  US09/141,289
• Time in Patent Office: 922 Days
• Field of Search: 382/128-133, 382/156, 382/159, 382/181, 382/190, 382/191, 382/195, 382/203, 382/224, 382/227, 382/291, 382/308, 378/37, 378/62, 378/63, 378/98.11, 600/410, 600/425, 600/408, 128/915
• US Class Current: 382/132
• CPC Class Codes: A61B 6/502 for diagnosis of breast, i.e. mammography
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06V 10/25 Determination of region of interest [ROI] or a volume of interest [VOI]
• CPC Class Codes: G06V 10/36 Applying a local operator, i.e. means to operate on image points situated in the vicinity of a given point; Non-linear local filtering operations, e.g. median filtering
• CPC Class Codes: G06V 10/457 by analysing connectivity, e.g. edge linking, connected component analysis or slices
• CPC Class Codes: G06V 10/48 by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

A method and system for detecting suspicious portions of digital mammograms by using independently calculated mass and spiculation information is disclosed. The method is for use in a computer aided diagnosis system that is designed to bring suspicious or possibly cancerous lesions in fibrous breast tissue to the attention of a radiologist or other medical professional. In a preferred embodiment, spiculation information and mass information are independently calculated, with the computed spiculation information not being dependent on results of the mass information computation, thus leading to greater reliability. Systems according to a preferred embodiment also compute spiculation information either prior to, or concurrently with, the computation of mass information, thus allowing increased overall system speed.

Claims

1. A method for automated detection of suspicious lesions in a digital mammogram, comprising the steps of:

computing mass information corresponding to the digital mammogram, said mass information having mass location information;

independent of said step of computing mass information, computing spiculation information corresponding to the digital mammogram, said spiculation information having spiculation location information; and

identifying suspicious lesions in the digital mammogram using said mass information and said spiculation information.

2. The method of claim 1, wherein said step of computing spiculation information is performed prior to or concurrently with said step of computing mass information.

3. The method of claim 2, wherein said spiculation information is computed from a set of information other than said mass information, for allowing faster and more reliable detection of suspicious lesions in the digital mammogram.

4. The method of claim 3, wherein said step of identifying suspicious lesions includes the step of using a classifier algorithm to classify a feature vector corresponding to at least one location in said digital mammogram, each feature vector comprising that portion of said mass information and said spiculation information that corresponds to that location.

5. The method of claim 4, wherein said classifier algorithm is used to classify feature vectors associated with each location in said digital mammogram.

6. The method of claim 4, wherein said classifier algorithm comprises a linear classifier algorithm.

7. The method of claim 4, wherein said classifier comprises a quadratic classifier algorithm.

8. The method of claim 6, said mass information comprising a mass metric, said spiculation information comprising a spiculation metric, said mass metric and said spiculation metric being scalar quantities, said linear classifier algorithm comprising the steps of:

weighting said mass metric by a first weight to produce a weighted mass metric;

weighting said spiculation metric by a second weight to produce a weighted spiculation metric;

combining said weighted mass metric and said weighted spiculation metric to produce a result;

comparing said result to a predetermined threshold;

identifying the corresponding location as suspicious if said result is greater than said predetermined threshold; and

identifying the corresponding location as normal if said result is not greater than said predetermined threshold.

9. The method of claim 8, wherein said first weight, said second weight, and said predetermined threshold are determined using a training algorithm on a training set of digital mammograms, said training set comprising a plurality of examples of suspicious lesions and a plurality of examples of normal breast structure.

10. A computer program product for directing a computing apparatus to automatically detect suspicious lesions in a digital mammogram, thus permitting the suspicious lesions to be brought to the attention of a medical professional, said computer program product comprising:

computer code for computing mass information corresponding to the digital mammogram, said mass information having mass location information;

computer code for computing, independent of the computation of said mass information, spiculation information corresponding to the digital mammogram, said spiculation information having spiculation location information; and

computer code for identifying suspicious lesions in the digital mammogram using said mass information and said spiculation information.

11. The computer program product of claim 10, wherein said spiculation information is computed prior to or concurrently with said step of computing mass information.

12. The computer program product of claim 11, wherein said computer code for identifying suspicious lesions includes computer code for using a classifier algorithm to classify a feature vector corresponding to at least one location in said digital mammogram, each feature vector comprising that portion of said mass information and said spiculation information that corresponds to that location.

13. The computer program product of claim 12, wherein said classifier algorithm is used to classify feature vectors associated with each location in said digital mammogram.

14. The computer program product of claim 13, said mass information comprising a mass metric, said spiculation information comprising a spiculation metric, said mass metric and said spiculation metric being scalar quantities, said linear classifier algorithm comprising the steps of:

weighting said mass metric by a first weight to produce a weighted mass metric;

weighting said spiculation metric by a second weight to produce a weighted spiculation metric;

combining said weighted mass metric and said weighted spiculation metric to produce a result;

comparing said result to a predetermined threshold;

identifying the corresponding location as suspicious if said result is greater than said predetermined threshold; and

identifying the corresponding location as normal if said result is not greater than said predetermined threshold.

15. An automated system for detecting suspicious portions of a digitized mammogram, comprising:

means for computing mass information corresponding to the digitized mammogram, said mass information having mass location information;

means for computing spiculation information corresponding to the digitized mammogram independent of said mass information, said spiculation information having spiculation location information; and

means for classifying said mass information and said spiculation information for detecting the suspicious portions of the digital mammogram.

16. The automated system of claim 15, wherein said means for computing spiculation information computes said spiculation information prior to or substantially in parallel with a period in which said means for computing mass information computes said mass information.

17. The automated system of claim 16, wherein said means for computing spiculation information computes said spiculation information using data other than said mass information.

18. The automated system of claim 17, wherein said means for classifying comprises:

means for computing at least one combined classification parameter defined by a combination of said mass information and said spiculation information; and

means for identifying values for which said at least one combined classification parameter corresponds to a suspicious portion of the digital mammogram.

19. The automated system of claim 17, wherein said means for classifying comprises means for implementing a neural network algorithm capable of identifying the suspicious portions of the digital mammogram using said mass information and said spiculation information.

20. The automated system of claim 18, wherein said means for classifying includes a look up table that is indexed according to scalar quantities associated with said mass information and said spiculation information.