Increased SNR and reduced distortions by averaging multiple gradient echo signals in 3D FLASH imaging of the human brain at 3T

Purpose

To demonstrate how averaging of multiple gradient echoes can improve high‐resolution FLASH (fast low angle shot) magnetic resonance imaging (MRI) of the human brain.

Materials and Methods

3D‐FLASH with multiple bipolar echoes was studied by simulation and in three experiments on human brain at 3T. First, the repetition time (TR) was increased by the square of the flip angle to maintain contrast as derived by theory. Then the number of echoes was increased at constant TR with bandwidths between 110 and 1370 Hz/pixel. Finally, signals of a 12‐echo acquisition train (echo times 4.9–59 msec) were averaged consecutively to study the increase in SNR.

Results

At unchanged contrast, the signal increased proportionally with flip angle and sqrt(TR). Increasing the bandwidth improved delineation of the basal cortex and vessels, while most of the loss in the signal‐to‐noise ratio (SNR) was recovered by averaging. Consecutive averaging increased the SNR to reach maximum efficiency at an echo train length corresponding roughly to T.

Conclusion

SNR is gained efficiently by acquiring additional echoes and increasing TR (and flip angle accordingly to maintain contrast) until the associated T loss in the averaged signal consumes the sqrt(TR) increase in the steady state. A bandwidth of 350 Hz/pixel or higher and echo trains shorter than T are recommended. J. Magn. Reson. Imaging 2009;29:198–204. © 2008 Wiley‐Liss, Inc.     

Advances of 3T MR imaging in visualizing trabecular bone structure of the calcaneus are partially SNR‐independent: Analysis using simulated noise in relation to micro‐CT, 1.5T MRI,and biomechanical strength

Purpose

To investigate differences in magnetic resonance imaging (MRI) of trabecular bone at 1.5T and 3.0T and to specifically study noise effects on the visualization and quantification of trabecular architecture using conventional histomorphometric and nonlinear measures of bone structure.

Materials and Methods

Sagittal MR images of 43 calcaneus specimens (donor age: 81 ± 10 years) were acquired at 1.5T and 3.0T using gradient echo sequences. Noise was added to obtain six sets of images with decreasing signal‐to‐noise ratios (SNRs). Micro‐CT images were obtained from biopsies taken from 37 calcaneus samples and bone strength was determined. Morphometric and nonlinear structure parameters were calculated in all datasets.

Results

Originally, SNR was 1.5 times higher at 3.0T. In the simulated image sets, SNR was similar at both fields. Trabecular dimensions measured by μCT were adequately estimated by MRI, with residual errors (e_r), ranging from 16% to 2.7% at 3.0T. Comparing e_r at similar SNR, 3.0T consistently displayed lower errors than 1.5T (eg, bone fraction at SNR ≈4: e_r[3.0T] = 15%; e_r[1.5T] = 21%, P < 0.05).

Conclusion

The advances of 3.0T compared to 1.5T in visualizing trabecular bone structure are partially SNR‐independent. The better performance at 3.0T may be explained by pronounced susceptibility, enhancing the visualization of thin trabecular structures. J. Magn. Reson. Imaging 2009;29:132–140. © 2008 Wiley‐Liss, Inc.     

Comprehensive quantification of signal‐to‐noise ratio and g‐factor for image‐based and k‐space‐based parallel imaging reconstructions

Parallel imaging reconstructions result in spatially varying noise amplification characterized by the g‐factor, precluding conventional measurements of noise from the final image. A simple Monte Carlo based method is proposed for all linear image reconstruction algorithms, which allows measurement of signal‐to‐noise ratio and g‐factor and is demonstrated for SENSE and GRAPPA reconstructions for accelerated acquisitions that have not previously been amenable to such assessment. Only a simple “prescan” measurement of noise amplitude and correlation in the phased‐array receiver, and a single accelerated image acquisition are required, allowing robust assessment of signal‐to‐noise ratio and g‐factor. The “pseudo multiple replica” method has been rigorously validated in phantoms and in vivo, showing excellent agreement with true multiple replica and analytical methods. This method is universally applicable to the parallel imaging reconstruction techniques used in clinical applications and will allow pixel‐by‐pixel image noise measurements for all parallel imaging strategies, allowing quantitative comparison between arbitrary k‐space trajectories, image reconstruction, or noise conditioning techniques. Magn Reson Med 60:895–907, 2008. © 2008 Wiley‐Liss, Inc.     

Optimizing the precision-per-unit-time of quantitative MR metrics: examples for T1, T2, and DTI.

Quantitative MR metrics (e.g., T1, T2, diffusion coefficients, and magnetization transfer ratios (MTRs etc)) are often derived from two images collected with one acquisition parameter changed between them (the "two-point" method). Since a low signal-to-noise-ratio (SNR) adversely affects the precision of these metrics, averaging is frequently used, although improvement accrues slowly-in proportion to the square root of imaging time. Fortunately, the relationship between the images' SNRs and the metric's precision can be exploited to our advantage. Using error propagation rules, we show that for a given sequence, specifying the total imaging time uniquely determines the optimal acquisition protocol. Specifically, instead of changing only one acquisition parameter and repeating the imaging pair until all available time is spent, we propose to adjust all of the parameters and the number of averages at each point according to their contribution to the sought metric's precision. The tactic is shown to increase the precision of the well-known two-point T1, T2, and diffusion coefficients estimation by 13-90% for the same sample, sequence, hardware, and duration. It is also shown that under this general framework, precision accrues faster than the square root of time. Tables of optimal parameters are provided for various experimental scenarios.     

MRI with zero echo time: Hard versus sweep pulse excitation

Zero echo time can be obtained in MRI by performing radiofrequency (RF) excitation as well as acquisition in the presence of a constant gradient applied for purely frequency‐encoded, radial centre‐out k‐space encoding. In this approach, the spatially nonselective excitation must uniformly cover the full frequency bandwidth spanned by the readout gradient. This can be accomplished either by short, hard RF pulses or by pulses with a frequency sweep as used in the SWIFT (Sweep imaging with Fourier transform) method for improved performance at limited RF amplitudes. In this work, the two options are compared with respect to T₂ sensitivity, signal‐to‐noise ratio (SNR), and SNR efficiency. In particular, the SNR implications of sweep excitation and of initial or periodical acquisition gaps required for transmit‐receive switching are investigated. It was found by simulations and experiments that, whereas equivalent in terms of T₂ sensitivity, the two techniques differ in SNR performance. With ideal, ungapped simultaneous excitation and acquisition, the sweep approach would yield higher SNR throughout due to larger feasible flip angles. However, acquisition gapping is found to take a significant SNR toll related to a reduced acquisition duty cycle, rendering hard pulse excitation superior for sufficient RF amplitude and also in the short‐T₂ limit. Magn Reson Med, 2011. © 2011 Wiley‐Liss, Inc.     

k-space interpretation of the Rose Model: noise limitation on the detectable resolution in MRI.

Noise limitation on the detected spatial resolution, described by the Rose Model, is well known in X-ray imaging and routinely used in designing X-ray imaging protocols. The purpose of this article is to revisit the Rose Model in the context of MRI where image data are acquired in the spatial frequency domain. A k-space signal-to-noise ratio (kSNR) is introduced to measure the relative signal and noise powers in a circular annulus in k-space. It is found that the kSNR diminishes rapidly with k-space radius. The Rose criterion that the voxel SNR approximately 4 is translated to kSNR cutoff values was tested using theoretical derivation and experimental histogram analysis. Experiments demonstrate that data acquisition beyond this cutoff k-space radius adds little or no information to the image. In order to reduce the noise limit on spatial resolution, the signal strength must be improved through means such as increasing the coil sensitivity, contrast enhancement, and signal averaging. This finding implies that the optimal k-space volume to be sampled or the optimal scan time in MRI should be matched to the relative SNR level.     

Prostate cancer detection with multi‐parametric MRI: Logistic regression analysis of quantitative T2, diffusion‐weighted imaging,and dynamic contrast‐enhanced MRI

Purpose

To develop a multi‐parametric model suitable for prospectively identifying prostate cancer in peripheral zone (PZ) using magnetic resonance imaging (MRI).

Materials and Methods

Twenty‐five radical prostatectomy patients (median age, 63 years; range, 44–72 years) had T2‐weighted, diffusion‐weighted imaging (DWI), T2‐mapping, and dynamic contrast‐enhanced (DCE) MRI at 1.5 Tesla (T) with endorectal coil to yield parameters apparent diffusion coefficient (ADC), T2, volume transfer constant (K^trans) and extravascular extracellular volume fraction (v_e). Whole‐mount histology was generated from surgical specimens and PZ tumors delineated. Thirty‐eight tumor outlines, one per tumor, and pathologically normal PZ regions were transferred to MR images. Receiver operating characteristic (ROC) curves were generated using all identified normal and tumor voxels. Step‐wise logistic‐regression modeling was performed, testing changes in deviance for significance. Areas under the ROC curves (A_z) were used to evaluate and compare performance.

Results

The best‐performing single‐parameter was ADC (mean A_z [95% confidence interval]: A_z,ADC: 0.689 [0.675, 0.702]; A_z,T2: 0.673 [0.659, 0.687]; A_z,Ktrans: 0.592 [0.578, 0.606]; A_z,ve: 0.543 [0.528, 0.557]). The optimal multi‐parametric model, LR‐3p, consisted of combining ADC, T2 and K^trans. Mean A_z,LR‐3p was 0.706 [0.692, 0.719], which was significantly higher than A_z,T2, A_z,Ktrans, and A_z,ve (P < 0.002). A_z,LR‐3p tended to be greater than A_z,ADC, however, this result was not statistically significant (P = 0.090).

Conclusion

Using logistic regression, an objective model capable of mapping PZ tumor with reasonable performance can be constructed. J. Magn. Reson. Imaging 2009;30:327–334. © 2009 Wiley‐Liss, Inc.