METHOD AND APPARATUS FOR SEGMENTATION AND REGISTRATION OF LONGITUDINAL IMAGES

• US 20140064580A1
• Filed: 01/09/2012
• Published: 03/06/2014
• Est. Priority Date: 01/11/2011
• Status: Active Grant

• Current Assignee: Rutgers University
• Original Assignee: Rutgers University
• Inventors: Madabhushi, Anant, Agner, Shannon C.
• Granted Patent: US 9,721,338 B2
• Time in Patent Office: Days
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10096 Dynamic contrast-enhanced magnetic resonance imaging [DCE-MRI]
• CPC Class Codes: G06T 2207/20116 Active contour; Active surface; Snakes
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/149 involving deformable models, e.g. active contour models
• CPC Class Codes: G06T 7/174 involving the use of two or more images
• CPC Class Codes: G06T 7/32 using correlation-based methods

Abstract

The described invention provides systems and methods for detecting and segmenting a lesion from longitudinal, time series, or multi-parametric imaging by utilizing spectral embedding-based active contour (SEAC). In addition, the described invention further provides systems and methods for registering time series data by utilizing reduced-dimension eigenvectors derived from spectral embedding (SE) of feature scenes (SERg).

Claims

1. A method for detecting and segmenting a lesion from longitudinal, time series, or multiparametric imaging by utilizing spectral embedding-based active contour (SEAC), the method comprising:

(a) obtaining a plurality of images from a longitudinal, time series, or multi-parametric imaging;

(b) applying spectral embedding (SE) to the longitudinal, time series, or multi-parametric imaging in a pixel-wise fashion to reduce a plurality of time point images to a single parametric embedding image (PrEIm), wherein the spectral embedding produces three smallest eigenvectors at each pixel location in the single parametric image by presenting color values in the parametric embedding image (PrEIm);

(c) calculating spatial tensor-based gradients on the parametric embedding image (PrEIm) derived from spectral embedding (SE) eigenvectors;

(d) selecting an initialization on the image;

(e) evolving active contour (AC) on the parametric embedding image (PrEIm); and

(f) detecting and segmenting the lesion based on morphological features derived from (e).

2. The method according to claim 1, wherein the longitudinal, time series, or multiparametric imaging comprises one or more of the following:

dynamic contrast enhanced magnetic resonance imaging (DCE-MRI);

T1 imaging;

T2 imaging; and

diffusion-weighted imaging.

6. The method according to claim 1, wherein at least one of steps (a), (b), (c), (d), (e) and (f) is performed using a computer.

7. The method according to claim 1, wherein at least one of steps (a), (b), (c), (d), (e) and (f) is performed automatically.

8. The method according to claim 1, wherein the lesion is a cancer.

9. The method according to claim 8, wherein the cancer is one or more of the following:

a breast cancer;

a prostate cancer;

a brain cancer;

a liver cancer;

a kidney cancer.

14. The method according to claim 1, wherein the lesion is a plaque in a blood vessel.

15. The method according to claim 14, wherein the blood vessel is a carotid artery.

3-5. -5. (canceled)

10-13. -13. (canceled)

16-27. -27. (canceled)

28. A system comprising:

a computer comprising a computer processor; and

a computer-readable storage medium tangibly storing thereon computer program instructions capable of being executed by the computer processor of the computing device, the computer program instructions defining steps of;

(a) obtaining a plurality of images from a longitudinal, time series, or multiparametric imaging;

(b) applying spectral embedding (SE) to longitudinal, time series, or multi-parametric imaging in a pixel-wise fashion to reduce a plurality of time point images to a single parametric embedding image (PrEIm), wherein the spectral embedding produces three smallest eigenvectors at each pixel location in the single parametric image by presenting color values in the parametric embedding image (PrEIm);

(c) calculating spatial tensor-based gradients on the parametric embedding image (PrEIm) derived from spectral embedding (SE) eigenvectors;

(d) selecting an initialization on the image;

(e) evolving active contour (AC) on the parametric embedding image (PrEIm); and

(f) detecting and segmenting a lesion based on morphological features derived from (e).

29. The system according to claim 28, wherein the longitudinal, time series, or multi-parametric imaging comprises one or more of the following:

dynamic contrast enhanced magnetic resonance imaging (DCE-MRI);

T1 imaging;

T2 imaging; and

diffusion-weighted imaging.

33. The system according to claim 28, wherein the lesion is a cancer.

34. The system according to claim 33, wherein the cancer is one or more of the following:

a breast cancer;

a prostate cancer;

a brain cancer;

a liver cancer;

a kidney cancer.

39. The system according to claim 28, wherein the lesion is a plaque in a blood vessel.

40. The system according to claim 39, wherein the blood vessel is a carotid artery.

30-32. -32. (canceled)

35-38. -38. (canceled)

41. A method for registering time series data by utilizing reduced-dimension eigenvectors derived from spectral embedding (SE) of a feature scene, the method comprising:

(a) obtaining a plurality of images from a longitudinal, time series, or multiparametric imaging;

(b) applying spectral embedding (SE) to time series data to transform a plurality of time series images into an alternative data presentation, wherein the spectral embedding (SE) places a pre-contrast image and a post-contrast image in a same embedding space and allows embedding eigenvectors to separate salient regions from non-salient regions in the image and to identify corresponding regions in the both pre-contrast and post-contrast images;

(c) driving registration based on areas of the salient regions and the corresponding regions; and

(d) capturing statistics of multi-dimensional data found in multiple spectral embedding (SE) eigenvectors.

42. The method according to claim 41, wherein capturing in (d) is performed using alphamutual information ([alpha]-MI).

43. (canceled)

44. (canceled)

45. A system comprising:

a computer comprising a computer processor; and

a computer-readable storage medium tangibly storing thereon computer program instructions capable of being executed by the computer processor of the computing device, the computer program instructions defining steps of;

(a) obtaining a plurality of images from a longitudinal, time series, or multiparametric imaging,

(b) applying spectral embedding (SE) to time series data to transform time series images into an alternative data presentation, wherein the spectral embedding (SE) places a pre-contrast image and a post-contrast image in a same embedding space and allows embedding eigenvectors to separate salient regions from non-salient regions in the image and to identify corresponding regions in the both pre-contrast and post-contrast images,

(c) driving registration based on areas of the salient regions and the corresponding regions, and

(d) capturing statistics of multi-dimensional data found in multiple spectral embedding eigenvectors.

46. The system according to claim 45, wherein capturing in (d) is performed using alphamutual information ([alpha]-MI).