BUDA‐MESMERISE: Rapid acquisition and unsupervised parameter estimation for T₁, T₂, M₀, B₀, and B₁ maps

Abstract

Purpose Rapid acquisition scheme and parameter estimation method are proposed to acquire distortion‐free spin‐ and stimulated‐echo signals and combine the signals with a physics‐driven unsupervised network to estimate T1, T2, and proton density (M0) parameter maps, along with B0 and B1 information from the acquired signals. Theory and Methods An imaging sequence with three 90° RF pulses is utilized to acquire spin‐ and stimulated‐echo signals. We utilize blip‐up/‐down acquisition to eliminate geometric distortion incurred by the effects of B0 inhomogeneity on rapid EPI acquisitions. For multislice imaging, echo‐shifting is applied to utilize dead time between the second and third RF pulses to encode information from additional slice positions. To estimate parameter maps from the spin‐ and stimulated‐echo signals with high fidelity, 2 estimation methods, analytic fitting and a novel unsupervised deep neural network method, are developed. Results The proposed acquisition provided distortion‐free T1, T2, relative proton density (M0), B0, and B1 maps with high fidelity both in phantom and in vivo brain experiments. From the rapidly acquired spin‐ and stimulated‐echo signals, analytic fitting and the network‐based method were able to estimate T1, T2, M0, B0, and B1 maps with high accuracy. Network estimates demonstrated noise robustness owing to the fact that the convolutional layers take information into account from spatially adjacent voxels. Conclusion The proposed acquisition/reconstruction technique enabled whole‐brain acquisition of coregistered, distortion‐free, T1, T2, M0, B0, and B1 maps at 1 × 1 × 5 mm3 resolution in 50 s. The proposed unsupervised neural network provided noise‐robust parameter estimates from this rapid acquisition.

FD‐Net: An unsupervised deep forward‐distortion model for susceptibility artifact correction in EPI

“…42 Physics-driven reconstruction of an echo-shifting acquisition has been proposed for relaxometry along with B 0 and B 1 mapping. 43 Note that these methods involve advanced pulse sequence modifications that may not be available at all sites, and often leverage TOPUP for estimation of field maps.…”

“…Self‐supervised denoising of a multicontrast multishot EPI sequence based on reversed‐PE acquisitions has been proposed for T, T * , and susceptibility mapping 42 . Physics‐driven reconstruction of an echo‐shifting acquisition has been proposed for relaxometry along with B and B mapping 43 . Note that these methods involve advanced pulse sequence modifications that may not be available at all sites, and often leverage TOPUP for estimation of field maps.…”

3D‐EPI blip‐up/down acquisition (BUDA) with CAIPI and joint Hankel structured low‐rank reconstruction for rapid distortion‐free high‐resolution T2* mapping

“…16,27 More recent work combined field map-informed parallel imaging with Hankel structured low-rank constraint to mitigate remaining phase errors, ghosts, and noise in distortion-free blip-up and -down acquisition (BUDA) EPI. 5,6,[28][29][30] Previous work on BUDA focused on 2D or simultaneous multislice (SMS) encoded imaging, 5,6,28 and did not explore 3D EPI readouts. 31 In this work, our contributions include the following:…”

“…Hybrid‐space SENSE incorporates this field map information into the multi‐shot SENSE model to obtain distortion‐free images 16,27 . More recent work combined field map‐informed parallel imaging with Hankel structured low‐rank constraint to mitigate remaining phase errors, ghosts, and noise in distortion‐free blip‐up and ‐down acquisition (BUDA) EPI 5,6,28–30 . Previous work on BUDA focused on 2D or simultaneous multislice (SMS) encoded imaging, 5,6,28 and did not explore 3D EPI readouts 31 …”

Intravoxel incoherent motion magnetic resonance imaging reconstruction from highly under-sampled diffusion-weighted PROPELLER acquisition data via physics-informed residual feedback unrolled network

“…It is significantly challenging to obtain a large number of IVIM parametric maps due to the long scan time. To reduce the dependence on realistic parametric maps, self-supervised learning uses the consistency between the estimated measurement from the physical model and the realistic measurement as a loss term in the deep neural network (Kaandorp et al 2021, Liu et al 2021, Huang 2022, So et al 2022. However, numerous realistic measured signals are still needed as training input data in self-supervised learning.…”