Three‐dimensional diffusion imaging with spiral encoded navigators from stimulated echoes (3D‐DISPENSE)

Abstract

Purpose To introduce a new method for motion‐insensitive 3D multishot diffusion imaging using 3D spiral‐encoded navigators from stimulated echoes (3D‐DISPENSE). Methods The 3D‐DISPENSE sequence contains a 3D stack‐of‐spiral navigator generated between the diffusion preparation and the turbo spin‐echo image acquisition from the twin pathway of a stimulated echo. Unlike normal navigator methods, 3D‐DISPENSE separates the navigator acquisition from the imaging readout without compromising the image SNR. Phase information from the navigators was included in an iterative image reconstruction algorithm to correct for intershot phase incoherence caused by motion. Results In a phantom experiment, 3D‐DISPENSE correctly estimated deliberately introduced phase errors. In a moving phantom, motion‐induced artifacts in the DWI were greatly mitigated by 3D‐DISPENSE. The ADC after 3D‐DISPENSE correction was identical to the reference. In a brain diffusion tensor experiment, phase‐incoherence artifacts from breathing, cardiac, and subject motion were removed almost perfectly in all view angles, resulting in distortion‐free DWI and color‐coded fractional anisotropy maps with 1.5‐mm isotropic resolution and nearly full brain coverage. Conclusion Three‐dimensional DISPENSE corrects motion‐induced phase‐incoherence artifacts in 3D multishot diffusion imaging and produces high‐quality 3D DWI and DTI.

Near‐silent distortionless DWI using magnetization‐prepared RUFIS

“…In multishot diffusion sequences, phase incoherence between each shot can be compensated by using the navigator methods . However, the navigator sequence cannot be used directly in the current sequence, as phase incoherence is converted into magnitude incoherence after the 90º tip‐back pulse, and this magnitude variation is difficult to separate from the main diffusion contrast.…”

“…In multishot diffusion sequences, phase incoherence between each shot can be compensated by using the navigator methods. 30,31 However, the navigator sequence cannot be used directly in the current sequence, as phase incoherence is converted into magnitude incoherence after the 90º tipback pulse, and this magnitude variation is difficult to separate from the main diffusion contrast. To minimize the error caused by the random magnitude incoherence, the current sequence is designed to use only 2 segments to acquire the center of the k-space, as shown in Figure 1B.…”

Spatial versus angular resolution for tractography-assisted planning of deep brain stimulation

“…Nevertheless, we chose for 7T to push for higher spatial resolutions to better evaluate the tradeoff between angular and spatial resolution. With ongoing developments in acceleration techniques such as multiband, or by using different readouts (Zhang et al., 2019), there is potential for higher spatial resolution acquisitions at 3T. Moreover, data processing packages have become better able to deal with distorted (Andersson and Sotiropoulos, 2015) and noisy data (Veraart et al., 2016), allowing to more readily push for higher spatial resolutions in diffusion acquisitions.…”

Distortion‐free 3D diffusion imaging of the prostate using a multishot diffusion‐prepared phase‐cycled acquisition and dictionary matching

“…Multishot sequences can also be combined with diffusion preparation (DP) modules (as opposed to DW), where the diffusion-sensitized transverse magnetization is stored into longitudinal magnetization by a 90° tip-up pulse. DP sequences have the advantage that diffusion encoding and readout are separated, and can thus be independently optimized for resolution, distortion, and signal-to-noise ratio (SNR), and can be combined with any readout scheme: turbo spin echo (TSE), 13,14 balanced steady-state free-precession (bSSFP), 15,16 and gradient echo (GRE). 17,18 In multishot diffusion-prepared (msDP) sequences, there are two main challenges that cause inaccurate apparent diffusion coefficient (ADC) quantification.…”

“…These T 1 effects make a simple monoexponential fitting unsuitable for accurate ADC quantification. 17,24 Possible solutions that have been suggested to account for these T 1 -related effects are: to increase TR or add a delay time between the shots to allow for full magnetization recovery, at the cost of increased acquisition time 17,24,25 ; to use a center-out trajectory, as the first readout segments are less T 1 -contaminated than later segments 18,19,24 ; to use phase cycling 25 ; or to use magnitude stabilizers, 13 which do not refocus unprepared magnetization during data acquisition.…”