Systems and methods for reducing respiratory-induced motion artifacts for accelerated imaging
In some aspects, the disclosed technology relates to reducing respiratory-induced motion artifacts for accelerated imaging. In one embodiment, magnetic resonance data may be acquired for an area of a subject containing the heart. The acquired data may include motion-corrupted data due to respiration of the subject. From the acquired data, an image may be independently reconstructed for each of a plurality of time frames, with each time frame corresponding to one of a plurality of heartbeats. A region containing the heart of the subject may be automatically detected in the reconstructed images, and rigid motion registration may be performed on the region of the reconstructed images containing the heart. Based on the rigid motion registration, a linear phase shift for motion correction may be determined. The linear phase shift may be applied to the motion-corrupted data to produce linear phase-shifted data, and a k-t image reconstruction may be performed on the linear phase-shifted data to produce motion-corrected images.
1. A method, comprising:
acquiring magnetic resonance data for an area of a subject containing the heart, wherein the acquired magnetic resonance data comprises motion-corrupted data due to respiration of the subject;
reconstructing, from the acquired magnetic resonance data, a preliminary image for each time frame of a plurality of time frames, wherein each time frame corresponds to a single heartbeat of a plurality of heartbeats of the subject and each respective preliminary image is reconstructed independently;
automatically detecting, in the preliminary reconstructed images, a region of interest that consists of a heart region of the subject and that excludes regions of the subject outside the heart region;
performing rigid motion registration on the region of interest in the preliminary reconstructed images to estimate displacement between respective reconstructed images, such that the heart region is aligned across a final reconstructed image series;
determining, based on the estimated displacement from the rigid motion registration, a linear phase shift for motion correction;
applying the linear phase shift to the acquired magnetic resonance data with the motion-corrupted data to produce linear phase-shifted data; and
performing a k-t image reconstruction on the linear phase-shifted data to produce a time series of images in which the region of interest is motion-corrected for the final reconstructed image series.
2. The method of claim 1 , wherein each respective image is reconstructed independently such as to avoid temporal blurring in images used for image registration.
3. The method of claim 1 , wherein reconstructing, from the acquired magnetic resonance data, a preliminary image for each time frame of a plurality of time frames uses parallel image reconstruction.
4. The method of claim 3 , wherein the parallel image reconstruction comprises iterative self-consistent parallel image reconstruction (SPIRiT).
5. The method of claim 1 , wherein performing rigid motion registration on the region of interest in the preliminary reconstructed images comprises performing registration in the image domain or k-space domain.
6. The method of claim 1 , wherein determining, based on the rigid registration, the linear phase shift comprises deriving linear k-space shifts in the x and y direction.
7. The method of claim 1 , wherein acquiring the magnetic resonance data comprises performing k-t accelerated acquisition.
8. The method of claim 1 , wherein performing the k-t image reconstruction on the linear phase-shifted data comprises k-t exploiting sparsity and low rank structure (k-t SLR), k-t principal component analysis (k-t PCA), or k-t PCA with sensitivity encoding (SENSE).
9. The method of claim 1 , wherein the magnetic resonance data is acquired during free-breathing of the subject or continuous data acquisition in multiple phases of the cardiac cycle of the heart of the subject.
10. The method of claim 1 , wherein the magnetic resonance data is acquired using a Cartesian, radial, or spiral k-space trajectory.
11. The method of claim 1 , wherein acquiring the magnetic resonance data comprises incoherent sampling and the k-t image reconstruction comprises compressed sensing reconstruction exploiting sparsity in the temporal direction.
12. A system, comprising:
a data acquisition device configured to acquire magnetic resonance data for an area of a subject containing the heart, wherein the acquired magnetic resonance data comprises motion- corrupted data due to respiration of the subject; and
one or more processors coupled to the data acquisition device and configured to cause the system to perform functions including:
reconstructing, from the acquired magnetic resonance data, a preliminary image for each time frame of a plurality of time frames, wherein each time frame corresponds to a single heartbeat of a plurality of heartbeats of the subject and each respective preliminary image is reconstructed independently;
automatically detecting, in the preliminary reconstructed images, a region of interest that consists of a heart region of the subject and that excludes regions of the subject outside the heart region;
performing rigid motion registration on the region of interest in the preliminary reconstructed images to estimate displacement between respective reconstructed images, such that the heart region is aligned across a final reconstructed image series;
determining, based on the estimated displacement from the rigid motion registration, a linear phase shift for motion correction;
applying the linear phase shift to the acquired magnetic resonance data with the motion-corrupted data to produce linear phase-shifted data; and
performing a k-t image reconstruction on the linear phase-shifted data to produce a time series of images in which the region of interest is motion-corrected for the final reconstructed time series.
13. The system of claim 12 , wherein each respective preliminary image is reconstructed independently such as to avoid temporal blurring in images used for image registration.
14. The system of claim 12 , wherein reconstructing, from the acquired magnetic resonance data, a reconstructed image for each time frame of a plurality of time frames uses parallel image reconstruction.
15. The system of claim 14 , wherein the parallel image reconstruction comprises iterative self-consistent parallel image reconstruction (SPIRiT).
16. The system of claim 12 , wherein performing rigid motion registration on the region of interest in the preliminary reconstructed images comprises performing registration in the image domain or k-space domain.
17. The system of claim 12 , wherein determining, based on the rigid registration, the linear phase shift comprises deriving linear k-space shifts in the x and y direction.
18. The system of claim 12 , wherein acquiring the magnetic resonance data comprises performing k-t accelerated acquisition.
19. The system of claim 12 , wherein performing the k-t image reconstruction on the linear phase-shifted data to produce motion corrected images comprises k-t exploiting sparsity and low rank structure (k-t SLR), k-t principal component analysis (k-t PCA), or k-t PCA with sensitivity encoding (SENSE).
20. The system of claim 12 , wherein the magnetic resonance data is acquired during free-breathing of the subject or continuous data acquisition in multiple phases of the cardiac cycle of the heart of the subject.
21. The system of claim 12 , wherein the magnetic resonance data is acquired using a Cartesian, radial, or spiral k-space trajectory.
22. The system of claim 12 , wherein acquiring the magnetic resonance data comprises incoherent sampling and the k-t image reconstruction comprises compressed sensing reconstruction exploiting sparsity in the temporal direction.
23. A non-transitory computer-readable medium having stored instructions that, when executed by one or more processors, cause a computing device to perform functions that comprise:
acquiring magnetic resonance data for an area of a subject containing the heart, wherein the acquired magnetic resonance data comprises motion-corrupted data due to respiration of the subject;
reconstructing, from the acquired magnetic resonance data, a preliminary image for each time frame of a plurality of time frames, wherein each time frame corresponds to a single heartbeat of a plurality of heartbeats of the subject and each respective preliminary image is reconstructed independently;
automatically detecting, in the preliminary reconstructed images, a region of interest that consists of a heart region of the subject and that excludes regions of the subject outside the heart region;
performing rigid motion registration on the region of interest in the preliminary reconstructed images to estimate displacement of the heart region between respective reconstructed images, such that the heart region is aligned across a final reconstructed image series;
determining, based on the estimated displacement from the rigid motion registration, a linear phase shift for motion correction;
applying the linear phase shift to the acquired magnetic resonance data with the motion- corrupted data to produce linear phase-shifted data; and
performing a k-t image reconstruction on the linear phase-shifted data to produce a time series of images in which the region of interest is motion-corrected images for the final reconstructed image series.