Automated 3D Reconstruction of the Cardiac Chambers from MRI and Ultrasound

Patent metadata

Field	Value
Publication number	US-2016140751-A1
Application number	US-201514929806-A
Country	US
Kind code	A1
Filing date	Nov 2, 2015
Priority date	Oct 31, 2014
Publication date	May 19, 2016
Grant date	—

Abstract

The disclosure relates to a method of automatically producing a three-dimensional (3D) segmentation of a heart chamber, the method comprising: obtaining data sets from cardiac magnetic resonance imaging (MRI) or ultrasound, generating a 3D segmentation of the heart chamber from the data sets using an active contour method, modifying the 3D segmentation by adding a plurality of intra-chamber structures; and identifying an enclosing myocardium using the 3D segmentation generated by the method.

First claim

1 . A method of automatically producing a three-dimensional (3D) segmentation of a heart chamber, the method comprising:
(a) obtaining data sets from cardiac magnetic resonance imaging (MRI) or ultrasound, (b) generating a 3D segmentation of the heart chamber from the data sets using an active contour method, (c) modifying the 3D segmentation by adding a plurality of intra-chamber structures; and (d) identifying an enclosing myocardium using the 3D segmentation generated in step (b).

2 . The method of claim 1 , where generating the 3D segmentation of the heart chamber from the MRI or ultrasound data sets comprises minimizing an energy function, E(Φ), when a contour lies on a boundary of the heart chamber, wherein E(Φ) is defined as
E (Φ)= E int (Φ)+ E ext (Φ),

wherein E int is the internal energy function and E ext is the external energy function of the heart chamber in a 3D domain.

3 . The method of claim 2 , where minimizing the energy function, E(Φ) comprises using an external energy function, E ext (Φ), defined as
E ext (Φ)= w 2 E reg +w 3 E edge +w 4 E geom

wherein E reg is a region-based term, E edge is an edge-based term, E geom is a geometric term, and where w 2 , w 3 , and w 4 are a plurality of weighting parameters.

4 . The method of claim 1 , where the MRI or ultrasound data sets comprise a plurality of short-axis cardiac magnetic resonance images, long-axis cardiac magnetic resonance images, sagittal MRI images, coronal MRI images, axial MRI images, or any combination thereof.

5 . The method of claim 3 , further comprising normalizing the MRI or ultrasound data sets and reusing the same weighting parameters across the entire MRI or ultrasound data set.

6 . The method of claim 1 , where modifying the 3D segmentation with a plurality of cardiac substructures comprises:
identifying a plurality of points on a convex hull of the 3D segmentation; computing a centroid for the plurality of points; calculating the radius and angle of the plurality of points on the convex hull with respect to the centroid to produce cylindrical coordinates for the plurality of points on the convex hull; and interpolating the cylindrical coordinates to produce a closed convex curve which includes the plurality of cardiac substructures.

7 . The method of claim 1 , where identifying an enclosing myocardium using the 3D segmentation comprises removing a portion of endocardium of the cardiac structure from the 3D segmentation and refilling the 3D estimation with a color pattern representing the myocardium of the cardiac structure in its place as the distance from the centroid is increased.

8 . The method of claim 1 , where generating a 3D segmentation of the cardiac structure from the MRI or ultrasound data sets comprises simultaneously segmenting the MRI or ultrasound data sets and reconstructing 3D images therefrom.

9 . The method of claim 1 , wherein said heart chamber is selected from the group consisting of the left ventricle, the right ventricle, the left atrium and the right atrium.

10 . The method of claim 1 , wherein said modifying the 3D segmentation by adding a plurality of intra-chamber structures comprises adding papillary muscles to a reconstructed volume.

11 . (canceled)

12 . The method of claim 1 , wherein a 3D contour of the heart chamber is non-covex, wherein a line connecting any two points inside the contour is not necessarily inside the contour, the method comprising identifying points on a convex hull of a contour, computing a centroid value by averaging over all the points, wherein the centroid point is used as a center of cylindrical coordinates and a radius and angle of all points on the convex hull are calculated based on a new coordinate system, wherein a new set of points constructs a closed convex curve that best approximates the non-convex contour.

13 . The method of claim 1 , further comprising extracting the enclosing myocardium from the rest of the 3D segmentation of the heart chamber.

14 . The method of claim 1 , further comprising calculating a volume of the heart chamber.

15 . A computer readable medium containing software instructions for preforming the method of claim 1 .

Assignees

• Univ California

Inventors

• Jafarkhani Hamid
• Hajiaghayi Mahdi
• Groves Elliott
• Kheradvar Arash

Classifications

• A61B8/5223

  for extracting a diagnostic or physiological parameter from medical diagnostic data (for algorithms to analyse biomedical images G06T7/0012) · CPC title

• G06T2207/30048

  Heart; Cardiac · CPC title

• G06T7/149

  involving deformable models, e.g. active contour models · CPC title

• A61B8/483

  involving the acquisition of a 3D volume of data · CPC title

• G06T2207/10132

  Ultrasound image · CPC title