Original contributionSignal-to-noise measures for magnetic resonance imagers
Abstract
The signal-to-noise ratio (SNR) in magnetic resonance imaging represents one of the system operating variables that must be determined both for evaluating the performance of different imaging protocols on a particular machine, and for monitoring machine performance as part of a routine quality control (QC) program. Utilizing a phantom and set of automated analysis programs currently under development, this study evaluated several ways of measuring image signal and noise and demonstrated the importance of utilizing measured voxel volumes as opposed to nominal volumes in the calculation of SNR. The NEMA proposed standard for SNR is compared with several other SNR measures and is recommended as the measure to be used in routine SNR reporting. The importance of utilizing other SNR measures in addition to the NEMA proposed standard for routine QC is discussed.
References (5)
- L. Kaufman et al.
Measuring signal-to-noise ratios in MR imaging
Radiology
(1989) - R.M. Sano
NEMA standards: Performance standards for clinical magnetic resonance systems
Cited by (70)
Diffusion tensor imaging within the healthy cervical spinal cord: Within- participants reliability and measurement error
2024, Magnetic Resonance ImagingDiffusion tensor imaging (DTI) is a promising technique for the visualization of the cervical spinal cord (CSC) in vivo. It provides information about the tissue structure of axonal white matter, and it is thought to be more sensitive than other MR imaging techniques for the evaluation of damage to tracts in the spinal cord.
The purpose of this study was to determine the within-participants reliability and error magnitude of measurements of DTI metrics in healthy human CSC.
A total of twenty healthy controls (10 male, mean age: 33.9 ± 3.5 years, 10 females, mean age: 47.5 ± 14.4 years), with no family history of any neurological disorders or a contraindication to MRI scanning were recruited over a period of two months. Each participant was scanned twice with an MRI 3 T scanner using standard DTI sequences. Spinal Cord Toolbox (SCT) software was used for image post-processing. Data were first corrected for motion artefact, then segmented, registered to a template, and then the DTI metrics were computed. The within-participants coefficients of variation (CV%), the single and average within-participants intraclass correlation coefficients (ICC) and Bland-Altman plots for WM, VC, DC and LC fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) were determined for the cervical spinal cord (between the 2nd and 5th cervical vertebrae).
DTI metrics showed poor to excellent within-participants reliability for both single and average ICC and moderate to high reproducibility for CV%, all variation dependent on the location of the ROI. The BA plots showed good within-participants agreement between the scan-rescan values.
Results from this reliability study demonstrate that clinical trials using the DTI technique are feasible and that DTI, in particular regions of the cord is suitable for use for the monitoring of degenerative WM changes.
Tailored magnetic resonance fingerprinting
2023, Magnetic Resonance ImagingNeuroimaging of certain pathologies requires both multi-parametric qualitative and quantitative imaging. The role of the quantitative MRI (qMRI) is well accepted but suffers from long acquisition times leading to patient discomfort, especially in geriatric and pediatric patients. Previous studies show that synthetic MRI can be used in order to reduce the scan time and provide qMRI as well as multi-contrast data. However, this approach suffers from artifacts such as partial volume and flow. In order to increase the scan efficiency (the number of contrasts and quantitative maps acquired per unit time), we designed, simulated, and demonstrated rapid, simultaneous, multi-contrast qualitative (T1 weighted, T1 fluid attenuated inversion recovery (FLAIR), T2 weighted, water, and fat), and quantitative imaging (T1 and T2 maps) through the approach of tailored MR fingerprinting (TMRF) to cover whole-brain in approximately four minutes.
We performed TMRF on in vivo four healthy human brains and in vitro ISMRM/NIST phantom and compared with vendor supplied gold standard (GS) and MRF sequences. All scans were performed on a 3 T GE Premier system and images were reconstructed offline using MATLAB. The reconstructed qualitative images were then subjected to custom DL denoising and gradient anisotropic diffusion denoising. The quantitative tissue parametric maps were reconstructed using a dense neural network to gain computational speed compared to dictionary matching. The grey matter and white matter tissues in qualitative and quantitative data for the in vivo datasets were segmented semi-automatically. The SNR and mean contrasts were plotted and compared across all three methods. The GS images show better SNR in all four subjects compared to MRF and TMRF (GS > TMRF>MRF). The T1 and T2 values of MRF are relatively overestimated as compared to GS and TMRF. The scan efficiency for TMRF is 1.72 min−1 which is higher compared to GS (0.32 min−1) and MRF (0.90 min−1).
A 3D personalized cardiac myocyte aggregate orientation model using MRI data-driven low-rank basis functions
2021, Medical Image AnalysisCardiac myocyte aggregate orientation has a strong impact on cardiac electrophysiology and mechanics. Studying the link between structural characteristics, strain, and stresses over the cardiac cycle and cardiac function requires a full volumetric representation of the microstructure. In this work, we exploit the structural similarity across hearts to extract a low-rank representation of predominant myocyte orientation in the left ventricle from high-resolution magnetic resonance ex-vivo cardiac diffusion tensor imaging (cDTI) in porcine hearts. We compared two reduction methods, Proper Generalized Decomposition combined with Singular Value Decomposition and Proper Orthogonal Decomposition. We demonstrate the existence of a general set of basis functions of aggregated myocyte orientation which defines a data-driven, personalizable, parametric model featuring higher flexibility than existing atlas and rule-based approaches. A more detailed representation of microstructure matching the available patient data can improve the accuracy of personalized computational models. Additionally, we approximate the myocyte orientation of one ex-vivo human heart and demonstrate the feasibility of transferring the basis functions to humans.
Head-and-Neck MRI-only radiotherapy treatment planning: From acquisition in treatment position to pseudo-CT generation
2020, Cancer/RadiotherapieIn context of head-and-neck radiotherapy, this study aims to compare MR image quality according to diagnostic (DIAG) and radiotherapy (RT) setups; and to optimise an MRI-protocol (including 3D T1 and T2-weighted sequences) for dose-planning (based on pseudo-CT generation).
To compare DIAG and RT setups, signal-to-noise-ratio (SNR) and percentage-image-uniformity (PIU) were computed on T1 images of phantoms and volunteers. Influence of the sample conductivity on SNR was quantified using homemade phantoms. To obtain reliable T1 and T2 images for RT-planning, an experimental design was performed on volunteers by using SNR, contrast-to-noise-ratio (CNR) and mean-opinion-score (MOS). Further, pseudo-CTs were generated from 8 patients T2 images with a state-of-art deep-learning method. These pseudo-CTs were evaluated by mean-absolute-error (MAE) and mean-error (ME).
SNR was higher for DIAG-setup compared to RT-setup (SNR-ratio = 1.3). A clear influence of the conductivity on SNR was observed. PIU was higher for DIAG-setup (38.8%) compared to RT-setup (33.5%). Regarding the protocol optimisation, SNR, CNR, and MOS were 20.6, 6.16, and 3.91 for the optimal T1 sequence. For the optimal T2 sequence, SNR, CNR and MOS were 25.6, 44.46 and 4.0. In the whole head-and-neck area, the mean MAE and ME of the pseudo-CTs were 82.8 and -3.9 HU.
We quantified the image quality decrease induces by using an RT-setup for head-and-neck radiotherapy. To compensate this decrease, an MRI protocol was optimised by using an experimental design. This protocol of 15 minutes provides accurate images which could be used for MRI-dose-planning in clinical practice.
Cette étude compare la qualité d’IRM obtenues à partir de systèmes d’acquisition dédiés au diagnostic (DIAG) et dédiés à la radiothérapie de la sphère ORL (RT); puis à optimiser un protocole d’IRM permettant le calcul de dose utilisant des pseudo-dvanographie qui a été validé sur huit patients.
Pour comparer les systèmes d’acquisition, le rapport-signal-sur-bruit et le pourcentage-d’uniformité-de-l’image (PIU) ont été calculés sur des images pondérées en T1 de fantômes et de volontaires. Afin d’obtenir des images adéquates pour la planification, un plan d’expérience a été réalisé. Par la suite, pour huit patients, des pseudo-scanographies ont été générées à partir d’une méthode d’apprentissage profond et évalués en utilisant l’erreur-moyenne-absolue (MAE) et l’erreur-moyenne (ME).
Le rapport-signal-sur-bruit était supérieur de 30% pour le système-DIAG comparé au système-RT. Le PIU était supérieur pour le système-DIAG (39%) comparé au système RT (33%). Le protocole optimisé comprenant deux séquences 3D dure 15 minutes. Avec ce protocole, les moyennes des MAE et ME des pseudo-scanographies générées sur toute la sphère ORL des patients, étaient de 82,8 et –3,9 UH (unités Hounsfield).
La perte en qualité d’images induite par l’utilisation d’un système-RT pour la radiothérapie ORL a été quantifiée. Pour compenser cette perte, un protocole d’IRM a été optimisé en utilisant un plan d’expérience. Ce protocole de 15 minutes permet de générer des pseudo-scanographies pour la planification de radiothérapie.
Characterizing Microstructural Tissue Properties in Multiple Sclerosis with Diffusion MRI at 7 T and 3 T: The Impact of the Experimental Design
2019, NeuroscienceCitation Excerpt :Diffusion data were pre-processed using FSL TOPUP and EDDY (Andersson et al., 2003; Andersson and Sotiropoulos, 2016) to correct for susceptibility-induced distortions, eddy currents and subject motion. Signal-to-Noise Ratio (SNR) values in the b0 images were calculated for each diffusion protocol in the whole brain using the difference method (Murphy et al., 1993), returning: SNR = 20.6 ± 2.2 for 3 T/low-b protocol, SNR = 15.1 ± 2.3 for 7 T/low-b protocol and SNR = 16.9 ± 1.7 for 3 T/high-b protocol. Boxplots of SNRs for the different protocols are shown in Fig. 1.
The recent introduction of advanced magnetic resonance (MR) imaging techniques to characterize focal and global degeneration in multiple sclerosis (MS), like the Composite Hindered and Restricted Model of Diffusion, or CHARMED, diffusional kurtosis imaging (DKI) and Neurite Orientation Dispersion and Density Imaging (NODDI) made available new tools to image axonal pathology non-invasively in vivo. These methods already showed greater sensitivity and specificity compared to conventional diffusion tensor-based metrics (e.g., fractional anisotropy), overcoming some of its limitations.
While previous studies uncovered global and focal axonal degeneration in MS patients compared to healthy controls, here our aim is to investigate and compare different diffusion MRI acquisition protocols in their ability to highlight microstructural differences between MS and control tissue over several much used models. For comparison, we contrasted the ability of fractional anisotropy measurements to uncover differences between lesion, normal-appearing white matter (WM), gray matter and healthy tissue under the same imaging protocols. We show that: (1) focal and diffuse differences in several microstructural parameters are observed under clinical settings; (2) advanced models (CHARMED, DKI and NODDI) have increased specificity and sensitivity to neurodegeneration when compared to fractional anisotropy measurements; and (3) both high (3 T) and ultra-high fields (7 T) are viable options for imaging tissue change in MS lesions and normal appearing WM, while higher b-values are less beneficial under the tested short-time (10 min acquisition) conditions.
T<inf>1</inf> relaxometry of crossing fibres in the human brain
2016, NeuroImageA comprehensive tract-based characterisation of white matter should include the ability to quantify myelin and axonal attributes irrespective of the complexity of fibre organisation within the voxel. Recently, a new experimental framework that combines inversion recovery and diffusion MRI, called inversion recovery diffusion tensor imaging (IR-DTI), was introduced and applied in an animal study. IR-DTI provides the ability to assign to each unique fibre population within a voxel a specific value of the longitudinal relaxation time, T1, which is a proxy for myelin content. Here, we apply the IR-DTI approach to the human brain in vivo on 7 healthy subjects for the first time. We demonstrate that the approach is able to measure differential tract properties in crossing fibre areas, reflecting the different myelination of tracts. We also show that tract-specific T1 has less inter-subject variability compared to conventional T1 in areas of crossing fibres, suggesting increased specificity to distinct fibre populations. Finally we show in simulations that changes in myelination selectively affecting one fibre bundle in crossing fibre areas can potentially be detected earlier using IR-DTI.