Evaluation of 3D fat-navigator based retrospective motion correction in the clinical setting of patients with brain tumors

Abstract

Purpose: A 3D fat-navigator (3D FatNavs) based retrospective motion correction is an elegant approach to correct for motion as it requires no additional hardware and can be acquired during existing 'dead-time' within common 3D protocols. The purpose of this study was to clinically evaluate 3D FatNavs in the work-up of brain tumors. Materials and Methods: An MRI-based fat-excitation motion navigator incorporated into a standard MPRAGE sequence was acquired in 40 consecutive patients with (or with suspected) brain tumors, pre and post-Gadolinium injection. Each case was categorized into key anatomical landmarks, the temporal lobes, the infra-tentorial region, the basal ganglia, the bifurcations of the middle cerebral artery and the A2 segment of the anterior cerebral artery. First, the severity of motion in the non-corrected MPRAGE was assessed for each landmark, using a 5-point score from 0 (no artifacts) to 4 (non-diagnostic). Second, the improvement in image quality in each pair and for each landmark was assessed blindly using a 4-point score from 0 (identical) to 3 (strong correction). Results: The mean image improvement score throughout the datasets was 0.54. Uncorrected cases with light and no artifacts displayed scores of 0.50 and 0.13 respectively, while cases with moderate artifacts, severe artifacts and non-diagnostic image quality revealed a mean score of 1.17, 2.25 and 1.38 respectively. Conclusion: Fat-navigator based retrospective motion correction significantly improved MPRAGE image quality in restless patients during MRI acquisition. There was no loss of image quality in patients with little or no motion, and improvements were consistent in patients who moved more.

Accuracy investigations for volumetric head‐motion navigators with and without EPI at 7 T

“…We chose to exclude movements with a rotation parameter > 10 • from our analysis, as we observed many outlying points for these big movements, which would drive the linear fitting, and therefore dominate the analysis unreasonable much compared with their relevance. Such big rotations are rarely seen, even with highly uncooperative patients, 21,22 and are also difficult to correct with retrospective corrections, as large gaps in k-space occur with sub-Nyquist sampling. 23 With prospective correction there is typically a limit to how large rotation angles can be corrected in real time, to not pose too conservative limits on the gradient strength and slew rate in the starting position.…”

Analysis of the effect of motion on highly accelerated 3D FatNavs in 3D brain images acquired at 3T

“…To reconstruct the missing lines, GRAPPA uses the auto-calibration signal (ACS) lines, which constitute the fully-sampled central region of a 3D FatNav volume collected prior to or during the main acquisition, with only a few seconds added to the scan total duration. Motion correction based on 3D FatNavs has been used to improve the image quality in MR images of the brain affected by non-deliberate motion [ 1 , 3 ] and by deliberate motion [ 4 , 5 ]. In [ 1 ], image sharpness was restored in high-resolution images by including the accelerated 3D FatNavs as part of an MP2RAGE and a TSE protocol with negligible increase in scanning time.…”

“…3D FatNavs have been tested in 40 patients with diagnosed or suspected brain tumors on a clinical 3T scanner (MAGNETOM Skyra 3T, Siemens Healthcare, Erlangen, Germany), resulting in visible improvements in image quality after motion correction [ 3 ], demonstrating to be a valuable tool for motion correction in clinical brain MRI. Moreover, its negligible additional scanning time and no need for extra hardware makes it likely to be preferable as a motion correction solution for less compliant subjects who might not tolerate long scans or needing to wear physical markers on their heads.…”

Tracking of rigid head motion duringMRIusing an EEGsystem

“…This may be needed to avoid artifacts from spin history effects resulting from through‐plane motion in 2D imaging, 5–7 local Nyquist violations in k‐space in multi‐shot sequences, 4 or inadequate coverage of the brain. However, some MR techniques in which excitation is nonselective and k‐space is sampled in 3 dimensions can be adequately compensated for motion through retrospective correction of samples in k‐space 8–11 . This typically relies on pose estimates to infer the actual path traveled in k‐space, used for re‐gridding of k‐space data 8,12 …”

“…However, some MR techniques in which excitation is nonselective and k-space is sampled in 3 dimensions can be adequately compensated for motion through retrospective correction of samples in k-space. [8][9][10][11] This typically relies on pose estimates to infer the actual path traveled in k-space, used for re-gridding of k-space data. 8,12 Most tracking methods can be categorized as either navigator-, optical-, or sensor-based.…”