Spatio-Temporal Free-Form Registration of Cardiac MR Image Sequences

Antonios Perperidiss009433610/02/2006

MR-Imaging Magnetic Resonance (MR) imaging allows

the acquisition of: 3D images which describe the cardiac anatomy. 4D cardiac image sequences which describe the

cardiac anatomy and function. Advances in cardiac MR imaging are

making it an important clinical tool: The improvement of the spatial and temporal

resolution of the image sequences enabling the imaging of small cardiac structures.

The development of tagged MR imaging which allow the study of cardiac motion.

Cardiac Image Registration Currently increased need for cardiac registration methods. Cardiac image registration is a very complex problem due to

4D nature of the cardiac data: Complicated non-rigid motion of the heart and the thorax. Low resolution with which cardiac images are usually acquired.

Recently, cardiac image registration has emerged as an important tool for a large number of applications such as: The construction of anatomical and functional atlases of the

heart. The analysis of the myocardial motion. The segmentation of cardiac images. The fusion of information from different modalities such as CT,

MR, PET, and SPECT. The comparison of images of the same subject. The comparisons between the cardiac anatomy and function of

different subjects.

Background There are currently many registration techniques

for cardiac imaging. Most techniques focus on 3D images ignoring any

temporal misalignment between two image sequences.

One exception: an approach for the spatial and temporal registration of cardiac SPECT and MR images. Uses linear interpolation for the temporal mapping

between the end-systolic and end-diastolic frames. The heart has a complicated motion pattern. This method ignores all the spatial information

contained in the images between the end-systolic and end-diastolic frames.

Spatio-Temporal Registration Introduction The heart is undergoing a spatially and

temporally varying degree motion during the cardiac cycle.

Spatial alignment of corresponding frames of the image sequences not enough since frames may not correspond to the same position in the cardiac cycle of the hearts.

This is due to differences in: The acquisition parameters. The length of cardiac cycles. The dynamic properties of the hearts .

Spatio-Temporal Registration Introduction

Spatio-temporal alignment enables to find the temporal relationship between the 2 image sequences.

We present 2 spatio-temporal alignment methods using image information only.

The 4-D mapping can be described by the transformation:

Mapping can be resolved into decoupled spatial and temporal components: Spatial: Temporal:

Spatial Alignment Aim: to relate each spatial point of an

image to a point of the reference image. Tspatial can be written: Tspatial/global: 3D affine transformation:

Coefficients parameterize the 12 degrees of freedom.

Tspatial/local: Free-Form deformation (FFD) model based on B-Splines.

Temporal Alignment

Ttemporal can be written: Ttemporal/global: affine transformation:

a accounts for scaling differences. b accounts for translation differences.

Ttemporal/local: Free-Form deformation (FFD) using 1-D B-Splines.

Combined Optimization of the Spatial and Temporal Components 2 registration algorithms for finding the optimal

transformation T:1. Optimizes the spatial and temporal transformation

components simultaneously using image information only 2. Optimizes the temporal transformation component before

optimizing the spatial component.

In the 1st algorithm: Optimal transformation T is found by maximizing a voxel based similarity measure.

Normalized Mutual Information (NMI): a measure of spatio-temporal alignment.

NMI is optimized as a function of Tspatial/global and Ttemporal/global using:

an iterative downhill descent algorithm. a simple iterative gradient descent method.

Separate Optimization of the Spatial and Temporal Components 1 Computational complexity of the

previous method is very high. Reduced by optimizing each

transformation component separately.

Ttemporal/global: aligns the temporal ends of the image sequences

Ttemporal/local: aligns a limited temporal positions of the cardiac cycles.

Temporal positions detected by calculating the normalized cross-correlation coefficient between each frame of the sequence with the first frame.

Separate Optimization of the Spatial and Temporal Components 2

The idea behind this approach: during the

contraction phase of the cardiac cycle each consecutive image will be less similar to the first image.

during the relaxation phase of the cardiac cycle each consecutive image will be more similar to the first image

Results Algorithm evaluation: 15 cardiac MR images from

healthy volunteers. Reference subject: 32 different time frames

acquired. 14 4D cardiac MR images were registered to the

reference subject. Cardiac Cycle length: 300 to 800msec. Qualitative evaluation through visual inspection. Quality of registration in spatial domain measured

by Calculating volume overlap for: The left and right ventricles. The myocardium.

Results – Separate optimization of the transformation components Maximum contraction & and-diastole positions

determined manually. Positions compared with corresponding positions

identified by the algorithm: Mean error in the detection of maximum contraction: 1.2

frames. Mean error of the end diastole detection: 0.93 frames.

Combined optimization of the transformation component - Qualitative Evaluation

Deformable temporal & spatial registration improves alignment of the image sequences in spatial and temporal domain

Combined optimization of the transformation component - Quantitative Evaluation

Optimising transformation components simultaneously provides better overlap measures the separate optimisation method.

Approach achieves very good spatio temporal registration of the images.

Computational complexity is reduced by 25% (approximately)

Results – Using cross correlation based method to calculate temporal alignment.

Applications of the Spatio-Temporal Registration Method. Large number of applications for this Spatio-

Temporal registration method: Comparison between image sequences from the

same subject. Comparison between image sequences from different

subjects, Building probabilistic

and statistical atlases of the cardiac anatomy and function.

Questions?