Method for analyzing detections in a set of digital images using case based normalcy classification

• US 6,763,128 B1
• Filed: 06/13/2003
• Issued: 07/13/2004
• Est. Priority Date: 06/23/1999
• Status: Expired due to Term

• Current Assignee: iCAD Incorporated
• Original Assignee: Qualia Computing Inc. (iCAD Incorporated)
• Inventors: Mitchell, Richard A., Collins, Michael J., Rogers, Steven K.
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): Akhavannik, Hussein
• Application Number:
  US10/461,316
• Time in Patent Office: 396 Days
• Field of Search: 382/130, 378/37
• US Class Current: 382/130
• CPC Class Codes: G06F 18/256 of results relating to different input data, e.g. multimodal recognition
• CPC Class Codes: G06T 2207/30068 Mammography; Breast
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• 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 10/443 by matching or filtering
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

A computer aided detection method and system to assist radiologists in the reading of medical images. The method and system has particular application to the area of mammography including detection of clustered microcalcifications and densities. A microcalcification detector is provided wherein individual detections are rank ordered and classified, and one of the features for classification is derived using a multilayer perceptron. A density detector is provided including an iterative, dynamic region growing module with embedded subsystem for rank ordering and classification of a best subset of candidate masks. A post processing stage is provided where detections are analyzed in the context of a set of images for a patient. The post processing includes a normalcy classification including providing computed values corresponding to each detection from a category of detections on an image set, computing a normalcy value using the computed values, and removing all detections from an image set when the normalcy value does not meet a predetermined condition. The final output of the system is a set of indications overlaid on the input medical images.

Claims

1. In a process for analyzing detections on a digital image set, a method for choosing or rejecting a category of detections on the image set including:

a) providing computed values corresponding to each detection from a category of detections on the image set wherein the detection categories comprise microcalcifications and densities;

b) computing a normalcy value using the computed values corresponding to the detections;

c) comparing the normalcy value to a predetermined threshold; and

d) rejecting all detections of the corresponding category if the normalcy value does not meet the predetermined threshold.

2. The method of claim 1 wherein the image set comprises a plurality of digital images and the detections are spread through the image set.

3. The method of claim 1 wherein a separate normalcy value is computed for each detection category, and the step of rejecting detections is performed separately on each detection category.

4. The method of claim 1 wherein the computed values corresponding to each detection comprise delta discriminant values.

5. The method of claim 4 wherein features are derived from the delta discriminant values for use in computing the normalcy value.

6. The method of claim 5 wherein the features derived from the delta discriminant values are input to a classifier to obtain the normalcy value.

7. The method of claim 6 wherein the classifier comprises a quadratic classifier.

8. The method of claim 5 wherein features derived from the delta discriminant values for each detection category are used to compute a separate normalcy value for each detection category, and the step of rejecting detections is performed separately on each detection category.

9. The method of claim 8 wherein the features derived from the delta discriminant values for microcalcifications comprise two features, the features being calculated by:

i) sorting microcalcification delta discriminant values, by image, for all images in the image set;

ii) identifying a within-image maximum delta discriminant value for each image;

iii) searching across all images in the image set and selecting the maximum delta discriminant value, to define a first feature;

iv) averaging the within image maximum delta discriminant values from all images in the image set, to define a second feature; and

said first and second features comprising inputs to a microcalcification normalcy classifier to obtain a microcalcification normalcy value for comparison to a predetermined microcalcification normalcy threshold value.

10. The method of claim 8 wherein a first and a second delta discriminant value is computed for each density detection and the features derived fro the delta discriminant values for densities comprise three features, the features being calculated by:

i) summing the first and second delta discriminants for each density detection;

ii) for each image, identifying a density detection having a maximum value resulting from the summing operation of step i) to define an identified detection for each image;

iii) computing the mean of the first and second delta discriminants for the identified detection on each image;

iv) selecting the minimum of the mean values computed in step iii), to define a first feature;

v) for each image, selecting the maximum delta discriminant from the first and second delta discriminants for the identified detection, to define a maximum delta discriminant value for each image;

vi) computing the mean of the maximum delta discriminant values for all the images, to define a second feature;

vii) for each image, selecting the minimum delta discriminant from the first and second delta discriminants for the identified detection, to define a minimum delta discriminant value for each image;

viii) computing the mean of the minimum delta discriminant values for all the images, to define a third feature;

the first, second and third features comprising inputs to a density classifier to obtain a density normalcy value for comparison to a predetermined density normalcy threshold value.

11. A system for choosing or rejecting a category of detections on the image set, the system comprising:

a) means for providing computed values corresponding to each detection from a category of detections on the image set wherein the detection categories comprise microcalcifications and densities;

b) means for computing a normalcy value using the computed values corresponding to the detections;

c) means for comparing the normalcy value to a predetermined threshold; and

d) means for rejecting all detections of the corresponding category if the normalcy value does not meet the predetermined threshold.

12. The system of claim 11 wherein the image set comprises a plurality of digital images and the detections are spread through the image set.

13. The system of claim 11 including means for computing a separate normalcy value for each detection category.

14. The system of claim 11 wherein the means for providing computed values corresponding to each detection provides delta discriminant values.

15. The system of claim 14 including means for deriving features from the delta discriminant values for use in computing the normalcy value.

16. The system of claim 15 including means to input the features derived from the delta discriminant values to a classifier to obtain the normalcy value.

17. The system of claim 16 wherein the classifier comprises a quadratic classifier.

18. The system of claim 16 wherein the means for deriving features derives features from the delta discriminant values for each detection category to compute a separate normalcy value for each detection category.

19. The system of claim 18 wherein the features derived from the delta discriminant values for microcalcifications comprise two features, and the means for deriving features comprises:

i) means for sorting microcalcification delta discriminant values, by image, for all images in the image set;

ii) means for identifying a within-image maximum delta discriminant value for each image;

iii) means for searching across all images in the image set and selecting the maximum delta discriminant value, to define a first feature;

iv) means for averaging the within image maximum delta discriminant values from all images in the image set, to define a second feature; and

v) a microcalcification normalcy classifier receiving the first and second features, the microcalcification normalcy classifier producing a microcalcification normalcy value for comparison to a predetermined microcalcification normalcy threshold value.

20. The system of claim 18 wherein a first and a second delta discriminant value is computed for each density detection and the features derived from the delta discriminant values for densities comprise three features, the features being calculated by:

i) means for summing the first and second delta discriminants for each density detection;

ii) means for identifying, for each image, a density detection having a maximum value resulting from the summing operation performed by the means for summing to define an identified detection for each image;

iii) means for computing the mean of the first and second delta discriminants for the identified detection on each image;

iv) means for selecting the minimum of the mean values computed by the means for computing the mean of the first and second delta discriminants, to define a first feature;

v) means for selecting, for each image, the maximum delta discriminant from the first and second delta discriminants for the identified detection, to define a maximum delta discriminant value for each image;

vi) means for computing the mean of the maximum delta discriminant values for all the images, to define a second feature;

vii) means for selecting, for each image, the minimum delta discriminant from the first and second delta discriminants for the identified detection, to define a minimum delta discriminant value for each image;

viii) means for computing the mean of the minimum delta discriminant values for all the images, to define a third feature; and

ix) a density normalcy classifier receiving the first, second and third features, the density normalcy classifier producing a density normalcy value for comparison to a predetermined density normalcy threshold value.