Automatic abnormal tissue detection in MRI images

• US 7,317,821 B2
• Filed: 11/22/2004
• Issued: 01/08/2008
• Est. Priority Date: 11/22/2004
• Status: Expired due to Fees

• Current Assignee: CareStream Health Incorporated
• Original Assignee: CareStream Health Incorporated
• Inventors: Chen, Shoupu, Ray, Lawrence A., Huo, Zhimin
• Primary Examiner(s): Ahmed; Samir
• Assistant Examiner(s): Tabatabai; Abolfazl
• Application Number:
  US10/994,715
• Publication Number:
  US 20060110018A1
• Time in Patent Office: 1,142 Days
• Field of Search: 382/128, 382/129, 382/130, 382/131, 382/132, 382/133, 382/134, 382/159, 382/164, 382/165, 382/171, 382/173, 382/177, 382/179, 382/224, 382/227, 600/409, 600/410, 600/524, 424/9.3
• US Class Current: 382/130
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06T 2207/30068 Mammography; Breast
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06V 2201/032 of protuberances, polyps nodules, etc.

Abstract

A pattern recognition method for automatic abnormal tissue detection and differentiation using contrast enhanced MRI images. The method includes the steps of acquiring a plurality of MRI breast image sets; aligning the plurality of MRI breast images with respect to spatial coordinates; differencing the plurality of MRI breast image sets with a reference MRI image set, producing a plurality of difference image sets; segmenting the plurality of difference image sets, producing a plurality of MRI breast images with segmented intensity pixels; applying dynamic system identification to the segmented intensity pixels, producing a plurality of dynamic system parameters; and classifying the plurality of system parameters augmented with other physical or non-physical factors into different classes.

Claims

1. A method for automatic material classification, comprising the steps of:

acquiring a plurality of image sets of an object sequentially in time;

aligning the plurality of image sets with respect to spatial coordinates;

differencing the plurality of image sets with a reference image set to produce a plurality of difference image sets;

segmenting the plurality of difference image sets to produce a plurality of images with segmented intensity pixels;

applying dynamic system identification to the segmented intensity pixels of the plurality of images to produce a plurality of dynamic system parameters; and

classifying the plurality of system parameters into different classes.

2. The method of claim 1, further comprising the step of, prior to classifying the plurality of system parameters, augmenting the system parameters with other physical or non-physical factors.

3. A method for abnormal tissue detection using contrast enhanced MRI images, comprising the steps of:

acquiring a plurality of MRI breast image sets sequentially in time;

aligning the plurality of MRI breast image sets with respect to spatial coordinates;

differencing the plurality of MRI breast image sets with a reference MRI image set to produce a plurality of difference image sets;

segmenting the plurality of difference image sets to produce a plurality of MRI breast image sets with segmented intensity pixels;

applying a dynamic system identification to the segmented intensity pixels of the plurality of MRI breast image sets to produce a plurality of dynamic system parameters; and

classifying the plurality of system parameters into different classes to detect abnormal tissue.

4. The method of claim 3, wherein the contrast enhanced MRI images are augmented with physical and/or non-physical factors.

5. The method of claim 3, further comprising the step of, prior to classifying the plurality of system parameters, augmenting the system parameters with other physical or non-physical factors.

6. The method of claim 3, wherein the step of acquiring a plurality of MRI breast image sets includes the steps of

acquiring a set of MRI images for a breast with a number of images in a spatial order prior to an injection of contrast agent;

acquiring a plurality of MRI image sets, each set having an equal number of images, of the same breast in the same spatial order after the injection of contrast agent; and

organizing the plurality of acquired MRI image sets in a temporal order.

7. The method of claim 3, wherein the step of aligning the plurality of MRI breast image sets with respect to spatial coordinates includes the steps of:

selecting one of the plurality of MRI image sets as an alignment reference set; and

using a localized registration to align the remaining plurality of MRI image sets to the alignment reference set.

8. The method of claim 3, wherein the step of differencing the plurality of MRI breast image sets with a reference MRI image set includes the steps of:

selecting one of the plurality of MRI image sets as an intensity reference image set; and

subtracting the intensity reference image set from the remaining plurality of MRI image sets to obtain the plurality of difference image sets.

9. The method of claim 3, wherein the step of segmenting the plurality of difference image sets includes the steps of

selecting a statistical intensity threshold;

thresholding the plurality of difference image sets using the statistical intensity threshold to set the pixels with intensity value under the threshold to zeros in the plurality of difference image sets; and

segmenting the plurality of MRI breast image sets according to the remaining non-zero pixels in the plurality of difference image sets to obtain the segmented intensity pixels in the plurality of MRI breast image sets.

10. The method of claim 3, wherein the step of applying a dynamic system identification includes the steps of:

(a) forming vectors of indexed images whereby indices are corresponding to a temporal order of the plurality of MRI breast image sets and each vector comprises the images at the same spatial position across the plurality of MRI breast image sets;

(b) in the formed vector of indexed images, determining an image that has a largest number of segmented intensity pixels and using the segmented intensity pixels as seed pixels;

(c) for each of the seed pixels, selecting corresponding segmented intensity pixels in the other images of the formed vector and forming an intensity pixel sequence, which includes the seed pixel, in the temporal order;

(d) applying dynamic system identification to the sequence of intensity pixels to produce the plurality of dynamic system parameters for the sequence of intensity pixels; and

(e) applying steps (b), (c), and (d) to the remaining vectors of indexed images.

11. The method of claim 10, wherein the step of classifying the plurality of system parameters includes steps of:

in a temporal order, collecting contrast enhanced MRI breast training image sets having different known abnormal tissues;

using the contrast enhanced MRI breast training image sets with known abnormal tissues to obtain sample dynamic system parameters for different abnormal tissues;

training a classifier using a supervised learning method using the sample dynamic system parameters augmented with other factors; and

classifying the dynamic system parameters augmented with other factors into different classes using the trained classifier.

12. A pattern recognition method for abnormal tissue detection using contrast enhanced MRI images, comprising the steps of:

acquiring a plurality of MRI breast image sets sequentially in time;

aligning the plurality of MRI breast image sets with respect to spatial coordinates;

generating a plurality of difference image sets by differencing the plurality of MRI breast image sets with a reference MRI image set;

generating a plurality of MRI breast image sets with segmented intensity pixels by segmenting the plurality of difference image sets;

producing a plurality of dynamic system parameters by applying a dynamic system identification to the segmented intensity pixels of the plurality of MRI breast image sets; and

classifying the plurality of system parameters into different classes to detect abnormal tissue.