Low-rank tensor imaging for multidimensional cardiovascular MRI

Patent metadata

Field	Value
Publication number	US-10436871-B2
Application number	US-201715495588-A
Country	US
Kind code	B2
Filing date	Apr 24, 2017
Priority date	Apr 24, 2017
Publication date	Oct 8, 2019
Grant date	Oct 8, 2019

Abstract

A new low rank tensor (LRT) imaging strategy/methodology, specifically for quantitative cardiovascular magnetic resonance (CMR) multitasking, includes performing a low-rank tensor image model exploiting image correlation along multiple physiological and physical time dimensions, a non-ECG data acquisition strategy featuring minimal gaps in acquisition and frequent collection of auxiliary subspace training data, and a factored tensor reconstruction approach which enforces the LRT model.

First claim

What is claimed is:

1. A method for performing magnetic resonance imaging (MRI) on a subject, comprising:
acquiring sparsely-sampled and spatially-encoded MRI imaging data for a region of interest in the subject;
obtaining a temporal factor tensor for the region of interest in the subject, the temporal factor tensor including at least one temporal basis function for each of one or more time-varying dimensions of the subject;
estimating a spatial factor matrix for the region of interest based on the sparsely sampled and spatially-encoded MRI imaging data and the temporal factor tensor, the spatial factor matrix including at least one spatial basis function for a spatially-varying dimension of the subject;
reconstructing a complete image for the region of interest by combining the spatial factor matrix and the temporal factor tensor.

2. The method of claim 1 , wherein the obtaining comprises:
acquiring training data at a subset of the spatial encodings for the region of interest;
calculating a training tensor representing the training data for the subset of spatial encodings; and
extracting the temporal factor tensor from the training tensor.

3. The method of claim 2 , wherein the extracting comprises decomposing the training tensor into a partially-encoded spatial factor matrix, a full core tensor, and temporal basis matrices, followed by calculation of the temporal factor tensor as a product of the core tensor and the temporal basis matrices.

4. The method of claim 2 , wherein the training data is acquired with only partial spatial encoding.

5. The method of claim 1 , wherein the temporal factor tensor comprises a product of a core tensor and one or more temporal basis matrices, each of the temporal basis matrices corresponding to a different time dimension.

6. The method of claim 5 , wherein the time dimension comprises one of cardiac phase, respiratory phase, elapsed time, imaging sequence parameters, or timing parameters.

7. The method of claim 1 , wherein the k-space locations for the training data correspond to k-space locations for identifying at least one of cardiac phases or respiratory phases for the subject.

8. The method of claim 1 , wherein the subject is a mammal.

9. The method of claim 1 , wherein the subject is a human.

10. A non-transitory machine-readable medium having machine executable instructions for causing one or more processors of a magnetic resonance imaging (MRI) machine to execute the imaging method of claim 1 .

11. The method of claim 1 , wherein the estimating comprises fitting the temporal factor tensor to the sparsely-sampled imaging data to obtain the spatial factor matrix.

12. The method of claim 1 , wherein the temporal factor tensor includes basis functions for only the one or more time-varying dimensions of the subject.

13. The method of claim 1 , wherein the temporal factor tensor and the spatial factor matrix are derived from separate data sets.

14. A magnetic resonance imaging (MRI) system, comprising:
a magnet operable to provide a magnetic field;
a transmitter operable to transmit to a region within the magnetic field;
a receiver operable to receive a magnetic resonance signal from the region; and
a processor operable to control the transmitter and the receiver; wherein the processor is configured to direct the transmitter and receiver to execute a sequence, comprising:
acquiring sparsely-sampled and spatially-encoded MRI imaging data for a region of interest in a subject;
obtaining a temporal factor tensor for the region of interest in the subject, the temporal factor tensor including at least one temporal basis function for each of one or more time-varying dimensions of the subject;
estimating a spatial factor matrix for the region of interest based on the sparsely sampled and spatially-encoded MRI imaging data and the temporal factor tensor, the spatial factor matrix including at least one spatial basis function for a spatially-varying dimension of the subject; and
reconstructing a complete image for the region of interest by combining the spatial factor matrix and the temporal factor tensor.

15. The MRI system of claim 14 , wherein the obtaining comprises:
acquiring training data at a subset of the spatial encodings for the region of interest;
calculating a training tensor representing the training data for the subset of spatial encodings; and
extracting the temporal factor tensor from the full training tensor.

16. The MRI system of claim 15 , wherein the extracting comprises decomposing the training tensor into a partially-encoded spatial factor matrix, a full core tensor, and full temporal basis matrices, followed by calculation of the temporal factor tensor as a product of the core tensor and temporal basis matrices.

17. The MRI system of claim 15 , wherein the training data is acquired with only partial spatial encoding.

18. The MRI system of claim 14 , wherein the estimating comprises fitting the temporal factor tensor to the sparsely sampled imaging data to obtain the spatial factor matrix.

19. The MRI system of claim 14 , wherein the temporal factor tensor comprises a product of a core tensor and one or more temporal basis matrices, each of the temporal basis matrices corresponding to a different time dimension.

20. The MRI system of claim 19 , wherein the time dimension comprises one of cardiac phase, respiratory phase, elapsed time, imaging sequence parameters, or timing parameters.

21. The MRI system of claim 14 , wherein the k-space locations for the training data correspond to k-space locations for identifying at least one of cardiac phases or respiratory phases for the subject.

22. The MRI system of claim 14 , wherein the subject is a mammal.

23. The MRI system of claim 14 , wherein the subject is a human.

24. The MRI system of claim 14 , wherein the temporal factor tensor includes basis functions for only the one or more time-varying dimensions of the subject.

25. The MRI system of claim 14 , wherein the temporal factor tensor and the spatial factor matrix are derived from separate data sets.

Assignees

• Cedars Sinai Medical Center

Inventors

• Li Debiao
• Christodoulou Anthony G
• Shaw Jaime
• Xie Yibin
• Nguyen Christopher

Classifications

• G01R33/56 Primary

  Image enhancement or correction, e.g. subtraction or averaging techniques {, e.g. improvement of signal-to-noise ratio and resolution} · CPC title

• G06V10/774

  Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06V10/7715 Primary

  Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/2136

  based on sparsity criteria, e.g. with an overcomplete basis · CPC title