Number 10-16: Improvement in Image Quality in Atrial Fibrillation by using a Novel Prospective Reconstruction Method
Case from: Moneal Shah1, Ann Robinson2, Scott Yang3, Vikas Rathi1. 1Center for Cardiovascular MRI, Allegheny General Hospital, Pittsburgh, PA 2Cardiac MRI, Methodist Hospital St. Louis Park, MN 3Cardiac MRI, Kaiser Permanente, Santa Rosa, CA
All the author's contributed equally to this case.
Background: Obtaining cardiac images using MRI is challenging due to the perpetual motion of the heart. In order to minimize motion artifacts, cardiac gating is performed to synchronize image acquisition with cardiac motion. The electrocardiogram and specifically the R wave are used to trigger image acquisition. A certain number of phases are acquired for each cardiac cycle. This technique is called prospective gating with retrospective reconstruction.
In atrial fibrillation (AF), due to the varying R-R intervals, the phase acquisition is not consistent between each cardiac cycle. This results in image blurring and often non-diagnostic images. Currently, there are few options to overcome this problem as described below:
1. When a patient has variable heart rates, one can set the heart rate to the approximate median heart rate with the thought that this will be the most representative cardiac cycle length. However, the faster and slower cycles will still add artifact to the images.
2. Parallel imaging can be used which allows for shorter breath hold and faster image acquisition that could minimize the amount of variability acquired.
3. Instead of ECG gating, pulse gating is another option but has not been found to be helpful either.
We present a novel method of image acquisition in patients with AF that we have found to significantly improve motion artifact introduced by variable heart rate. This is achieved by performing prospective reconstruction as described below:
1. Take the highest heart rate of the patient and therefore the shortest R-R interval and set this as the specified heart rate.
2. Instead of prescribing the number of cardiac phases, we use the automated option on the scanner. By doing this, the k-space is filled prospectively with R-R intervals only as long as the shortest interval. If an R-R interval is longer than the pre-specified interval, the remaining portion of the cardiac cycle will be truncated and will not be included in the k-space. The reasoning behind this is that the duration of systole is relatively constant regardless of the heart rate variability.
3. There will be some data loss from the truncated segment, however, this corresponds to the last portion of diastole. By applying this new method in a patient with AF, it makes each cardiac cycle more uniform. The following examples show the difference in image quality between retrospective and prospective reconstruction (Movie 1 and 2 respectively). Compared to the image acquired by retrospective reconstruction, the prospectively reconstructed image demonstrates no image blurring, improved blood endocardial border definition and improved visualization of the valves and trabeculations.
Movie 1: Retrospective (right click to enlarge) Movie 2: Prospective (right click to enlarge)
The improved image quality using prospective reconstruction is also seen in patients with AF but with low heart rate variability (Movie 3 and 4). Movie 4 shows better detail of the pericardium (no blurring at the apex) and improved visualization of the endocardial borders and tricuspid regurgitatant jet.
Movie 3: Retrospective (right click to enlarge) Movie 4: Prospective (right click to enlarge)
More examples of standard SSFP images comparing the retrospective vs. prospective reconstruction are shown below demonstrating the improved resolution of the aortic valve leaflets (Movie 5 vs. Movie 6).
Movie 5: Retrospective (right click to enlarge) Movie 6: Prospective (right click to enlarge)
The prospective reconstruction in AF patients has demonstrable utility not only in standard SSFP images but also in tagging (Movie 7).
Movie 7 (right click to enlarge)
Limitations: Due to the use of a fixed short cycle length acquisition, the terminal portion of the diastolic period is truncated. This leads to images which have complete systolic information but partial diastolic information. Therefore, these images cannot be used for diastolic volumes and diastolic function evaluation.