2D CAIPI accelerated 3D multi-slab diffusion weighted EPI combined with qModeL reconstruction for fast high resolution microstructure imaging

Abstract

Purpose: To develop acceleration strategies for 3D multi-slab diffusion weighted imaging (3D ms-DWI) for enabling applications that require simultaneously high spatial (1 mm isotropic) and angular (> 30 directions) resolutions.

Methods: 3D ms-DWI offers high SNR-efficiency, with the ability to achieve high isotropic spatial resolution, yet suffers from long scan-times for studies requiring high angular resolutions. We develop 6D k-q space acceleration strategies to reduce the scan-time. Specifically, we develop non-uniform 3D k_y-k_z under-sampling employing a shot-selective 2D CAIPI sampling approach. To achieve inter-shot phase-compensation, 2D navigators were employed that utilize the same CAIPI trajectory. An iterative model-based 3D multi-shot reconstruction was designed by incorporating phase into the forward encoding process. Additionally, the shot-selective non-uniform k_y-k_z CAIPI acceleration was randomized along the q-dimension. The 3D model-based multi-shot reconstruction is then extended to a joint reconstruction that simultaneously reconstructs all the q-space points, with the help of a spatial total variation and deep-learned q-space regularization.

Results: The proposed reconstruction is shown to achieve adequate phase-compensation in both 2D CAIPI accelerated and additional k_y-k_z under-sampled cases. Using retrospective under-sampling experiments, we show that k-q accelerations close a factor of 12 can be achieved with a reconstruction error < 3% for both single and multi-shell data. This translates to a scan-time reduction by 3-fold for experiments with simultaneously high spatial and angular resolutions.

Conclusion: The proposed method facilitates the utilization of 3D ms-DWI for simultaneously high k-q resolution applications with close to 3× reduced scan-time.

Keywords: 3D DWI; CAIPI; K-q acceleration; Kurtosis; Multi-slab; Over-sampling; qModeL.

Figure 1:

3D-EPI multi-shot acquisition using shot-selective CAIPI: Here (A)-(B) represents a fully sampled case which employ 4 k_y shots (denoted by four colors) to sample every k_z plane. (B) is a short-hand representation of (A), where the k_x axis points into the page and is not displayed. Only a few points on the k_y axis are displayed for simplicity. Note that since each k_y shot and each k_z plane corresponds to a TR, the fully sampled case illustrated in (A)-(B) is time-consuming. CAIPI acceleration schemes can be used to reduce the scan time. (C) Using the shot-selective CAIPI method, CAIPI acceleration by a factor of 4 is illustrated, where each k_z plane employs only one k_y shot, out of the 4 shots. (D) shows the case that leads to further acceleration where some k_z planes are fully skipped, which involves fewer TRs than (C). The color-filled circles denote the sampled k-space locations and the hollow circles show the k-space locations that are not-sampled. Here, N_slices is the total number of k_z encodings needed for reconstruction at full resolution (accounting for over-sampling) and N_sh is the number of k_z shots acquired in the accelerated case.

Figure 2:

3D under-sampled reconstruction for a case with no phase inconsistency. The 2D CG-SENSE reconstruction are shown for: (A) the fully sampled case, (B): the 2D CAIPI acceleration (case in Figure 1C), and (C): an additional 4-fold under-sampled case (case in Figure 1D). The difference map for the accelerated cases in (B), (C) scaled by a factor of 5 is shown in (D), (E) respectively. (F)-(G) shows the slices corresponding to cases in (A),(C) respectively. (H) shows the plot of the reconstruction errors at various acceleration factors. (I) shows the plot of the SNR computed at various accelerations.

Figure 3:

(A) - (B) show the schematic of the 3D multi-shot forward encoding process for a given shot j , and for all shots, respectively. (C) shows the steps involved in the computation of the composite sensitivity maps.

Figure 4:

Phase compensated recovery with 2D CAIPI acceleration (R_y=4.28, R_z=1): The slice locations from a 2D single-shot DW-EPI are shown in (A) and (D) for the dataset 1 and dataset 3 respectively as a reference. (B) and (E) show the 3D multi-shot independent reconstruction with phase-compensation for the CAIPI accelerated case. (C) and (F) shows the proposed phase-compensated joint reconstruction with q-space regularization. While the joint reconstruction involves reconstructing all the q-space points jointly, the independent reconstructs one q-space data at a time. A zoomed view of the slices are given in supporting information.

Figure 5:

(A)-(E) show k_y-k_z shot-accelerated joint reconstruction for dataset 1 corresponding to acceleration factors of R_z = 2.5 and 3.33 for a given diffusion direction. Here, the non-uniform and randomized k-q CAIPI sampling pattern is utilized. (F)-(G) show the 3D multi-shot independent recovery for the same diffusion direction.

Figure 6:

k_y-k_z shot-accelerated joint reconstruction for dataset 1 corresponding to acceleration factors of R_z = 2.5 and 3.33 are shown for a given diffusion direction. Here, the same uniform CAIPI sampling pattern is utilized for all q-space points.

Figure 7:

Under-sampled reconstructions of dataset 3 at various acceleration factors. (A)-(B) shows the reconstructed DWIs from a given slab. The DWIs from b1000 and b2000 shells are displayed, which show adequate phase compensation. (C) -(D) show the error maps w.r.t R_z =1. The error maps are scaled by a factor of 2.

Figure 8:

(A)-(C) show the fractional anisotropy maps computed from the diffusion kurtosis model shown for dataset 3 at various under-sampling factors. (D)-(E) show the error maps w.r.t R_z =1. All images are displayed on the same scale from [0 1].