We have located links that may give you full text access.
Reducing Clustering of Readouts in Non-Cartesian CINE MRI.
Journal of Cardiovascular Magnetic Resonance 2024 January 29
BACKGROUND: The use of non-Cartesian MRI trajectories at golden angle increments has the advantage of allowing retrospective motion correction and image-based gating using intermediate real-time reconstructions. However, when the acquired data is cardiac binned for CINE imaging, trajectories are found to cluster together at certain heart rates and leave large unsampled gaps in k-space, leading to image artifacts. In this work, we (1) demonstrate an approach to reduce clustering by inserting additional angular rotations periodically within the trajectory, and (2) optimize for these additional angles using particle swarm optimization while still allowing for important intermediate reconstructions.
METHODS: Three acquisition models were simulated under constant and variable heart rates: traditional golden angle (Mtrd ), random additional angles (Mrnd ), and optimized additional angles (Mopt ). To analyze clustering, the standard deviation of the trajectory angular differences (STAD) was computed. The resulting distributions of STAD were compared through their interquartile ranges and the Kolmogorov-Smirnov test (significance level: p = 0.05). Agreement between a reference image reconstructed with uniform sampling and images obtained from Mtrd , Mrnd , and Mopt was analyzed by computing the structural similarity index measure (SSIM) and its interquartile range. Mtrd and Mopt were then compared in 3 healthy adults at 3 levels of heart rate variability (high, low and none).
RESULTS: The STAD distributions from Mtrd were significantly different (p < 0.05) from those with the Mopt and Mrnd . The STAD (interquartile range x 10-2 rad) showed that clustering was reduced with Mopt (0.5) and Mrnd (0.5) when compared to Mtrd (1.9) at constant heart rates. Similarly for variable heart rates, Mopt (0.5) and Mrnd (0.5) outperformed Mtrd (0.9). Risk for clustering was reduced with the proposed approach. The SSIM (interquartile range), relative to a ground truth reconstruction, showed that the best image quality was produced by Mopt (0.011), followed by Mrnd (0.014), and Mtrd (0.030), which produced the worst image quality. In-vivo studies showed that at reduced heart variability Mopt outperformed Mtrd with reduced risk for clustering. In high heartrate variability, both models performed well.
CONCLUSION: This approach reduces occurrences of clustering in k-space and improves resulting image quality without affecting acquisition time.
METHODS: Three acquisition models were simulated under constant and variable heart rates: traditional golden angle (Mtrd ), random additional angles (Mrnd ), and optimized additional angles (Mopt ). To analyze clustering, the standard deviation of the trajectory angular differences (STAD) was computed. The resulting distributions of STAD were compared through their interquartile ranges and the Kolmogorov-Smirnov test (significance level: p = 0.05). Agreement between a reference image reconstructed with uniform sampling and images obtained from Mtrd , Mrnd , and Mopt was analyzed by computing the structural similarity index measure (SSIM) and its interquartile range. Mtrd and Mopt were then compared in 3 healthy adults at 3 levels of heart rate variability (high, low and none).
RESULTS: The STAD distributions from Mtrd were significantly different (p < 0.05) from those with the Mopt and Mrnd . The STAD (interquartile range x 10-2 rad) showed that clustering was reduced with Mopt (0.5) and Mrnd (0.5) when compared to Mtrd (1.9) at constant heart rates. Similarly for variable heart rates, Mopt (0.5) and Mrnd (0.5) outperformed Mtrd (0.9). Risk for clustering was reduced with the proposed approach. The SSIM (interquartile range), relative to a ground truth reconstruction, showed that the best image quality was produced by Mopt (0.011), followed by Mrnd (0.014), and Mtrd (0.030), which produced the worst image quality. In-vivo studies showed that at reduced heart variability Mopt outperformed Mtrd with reduced risk for clustering. In high heartrate variability, both models performed well.
CONCLUSION: This approach reduces occurrences of clustering in k-space and improves resulting image quality without affecting acquisition time.
Full text links
Related Resources
Get seemless 1-tap access through your institution/university
For the best experience, use the Read mobile app
All material on this website is protected by copyright, Copyright © 1994-2024 by WebMD LLC.
This website also contains material copyrighted by 3rd parties.
By using this service, you agree to our terms of use and privacy policy.
Your Privacy Choices
You can now claim free CME credits for this literature searchClaim now
Get seemless 1-tap access through your institution/university
For the best experience, use the Read mobile app