21–26 Jun 2026
U. Ottawa - Learning Crossroads (CRX) Building
America/Toronto timezone
Welcome to the 2026 CAP Congress Program website! / Bienvenue au siteweb du programme du Congrès de l'ACP 2026!

Acceleration of the MP2RAGE MRI Pulse Sequence Through Multi-Contrast Joint Reconstruction

23 Jun 2026, 15:00
15m
U. Ottawa - Learning Crossroads (CRX) Building

U. Ottawa - Learning Crossroads (CRX) Building

100 Louis-Pasteur Private, Ottawa, ON K1N 9N3
Oral not-in-competition (Graduate Student) / Orale non-compétitive (Étudiant(e) du 2e ou 3e cycle) Physics in Medicine and Biology / Physique en médecine et en biologie (DPMB-DPMB) (DPMB) T2-2 | (DPMB)

Speaker

Curtis Goosney (McMaster University)

Description

The Magnetization-Prepared Two Rapid Gradient Echo (MP2RAGE) pulse sequence acquires two high-resolution 3D volumes that are combined to produce a T1-weighted UNI image and a quantitative T1 map that is largely insensitive to B1+ inhomogeneity. In the brain, T1 relaxation is strongly correlated with myelin content, enabling in vivo assessment of neurological health. However, MP2RAGE scan times remain relatively long (~7–10 min) when using conventional GRAPPA reconstruction with reduction factors (R) of 2–3. Multi-contrast joint reconstruction methods such as J-GRAPPA, JVC-GRAPPA, and J-LORAKS have demonstrated improved performance over GRAPPA for other pulse sequences by leveraging shared k-space information across contrasts. In this work, we investigate the application of joint reconstruction to MP2RAGE by exploiting shared information between the two acquired volumes to enable higher undersampling without compromising UNI image quality or T1 map precision. Fully sampled 1-mm isotropic MP2RAGE scans were performed on two healthy adult participants and retrospectively undersampled using uniform Cartesian masks for all algorithms and, additionally, variable-density Poisson-disc (VDPD) masks for J-LORAKS. Reconstruction performance was evaluated using root-mean-square error (RMSE) relative to fully sampled references while varying R and the number of autocalibration signal lines (NACS).
As expected, T1 map RMSE trends closely followed those of the UNI images. For T1 maps, JVC-GRAPPA and J-LORAKS consistently outperformed GRAPPA at higher reduction factors, achieving up to ~40% RMSE reduction with uniform sampling at R = 6 with NACS = 48. Using VDPD sampling with J-LORAKS further reduced T1 map RMSE by 18% at Rnet = 3.35 (R/NACS = 6/48) compared to uniform sampling and eliminated visible coherent aliasing. Reconstructions at higher reduction factors of R = 8-10 with J-LORAKS (Rnet = 3.80-4.13) yielded a ~68% and ~30% improvement with a VDPD sampling mask as compared to GRAPPA and JVC-GRAPPA with a uniform Cartesian sampling mask, respectively. Joint reconstruction methods also produced gray and white-matter T1 estimates closer to the fully sampled reference (p < 0.05). These results demonstrate the feasibility and of multi-contrast joint reconstruction for further accelerating MP2RAGE acquisitions while maintaining image quality and T1 precision.

Keyword-1 Medical Imaging
Keyword-2 MRI
Keyword-3 Quantitative Mapping

Author

Curtis Goosney (McMaster University)

Co-author

Chris Rowley (Physics & Astronomy, McMaster University)

Presentation materials

There are no materials yet.