High-resolution magnetic resonance imaging. Technical concepts and their implementation

Qualitative and quantitative analysis of routinely postprocessed (CLEAR) CE-MRA data sets: are SNR and CNR calculations reliable?

RATIONALE AND OBJECTIVES

To evaluate objective image quality parameters for contrast-enhanced magnetic resonance angiography (CE-MRA), contrast-to-noise (CNR), and signal-to-noise ratio (SNR) calculations based on signal intensity (SI) and standard deviation (SD) measurements of the vessel, the surrounding tissue (eg, muscle), and the background noise outside the body are commonly used. However, modern magnetic resonance scanners often use dedicated software algorithms such as Constant LEvel AppeaRance (CLEAR) to improve image quality, which may affect the established methods of SNR and CNR calculation. The purpose of this study was to intraindividually evaluate the feasibility of conventional techniques used for SNR and CNR calculation of MRA data sets that have been reconstructed with both, a standard (non-CLEAR) and a CLEAR algorithm.

METHODS

Supra-aortic high-resolution CE-MRA of 11 patients with headache symptoms was performed at 1.5 T using reconstruction algorithms generating both, non-CLEAR and CLEAR-corrected images from the acquired data set. A qualitative analysis with regard to image quality and contrast level was performed by two radiologists applying a score system. For quantitative analysis, distribution of SI values was measured in regions of interest in the common carotid artery (CCA) and the C1 segment of the internal carotid artery in identical positions of both data sets for intraindividual comparison of SNR and CNR calculations. For that purpose, three different equations were used for background noise assessment by determining the SD of SIs measured in the air outside the body (Eq. A), the soft tissue adjacent to the analyzed vessel segment (Eq. B), and in a contrast-medium filled tube (reference standard), which was placed around the patient's neck (Eq. C).

RESULTS

The qualitative analysis documented an improved image quality and a higher contrast level for CLEAR-based data sets. SNR and CNR calculations of the CCA and the C1 segment were significantly different for both reconstruction algorithms when using the background noise outside the body for image noise assessment (P<.05 [CCA]; P<.05 [C1]). SNR and CNR calculations based on the soft tissue adjacent to the analyzed segment or a reference standard were comparable.

CONCLUSIONS

For comparative analysis of CE-MRA data sets, SNR and CNR calculations based on SD determination of the background noise signal measured outside the body are not applicable for CE-MRA data sets reconstructed with a CLEAR-based algorithm. Therefore, noise should rather be assessed in the perivascular tissue to enable proper comparative analysis of CLEAR-enhanced CE-MRA data sets.

Optimizing lesion contrast without using contrast agents

The essence of optimizing lesion detection is increasing the signal-to-noise (S/N) for the lesion while decreasing the S/N of the background. Maneuvers which improve one aspect of lesion detection may have a deleterious effect on another. For example, thinner slices decrease partial volume averaging (important for detection of small lesions) but also decrease S/N. Attempts to increase S/N by lowering the bandwidth may increase the TE, decreasing the degree of T1-weighting. Fast spin echo (FSE) generally offers the best T2-weighting for detection of long T2 lesions. However, lesions with short T2s (due to magnetic susceptibility effects) are better detected with gradient echo techniques. Strategies which decrease background S/N include those based on chemical shift differences (ie, spectroscopic techniques like FatSat) and those based on differences in T1 (ie, inversion recovery [IR] techniques like STIR and FLAIR).

Contrast-enhanced MR imaging of the liver: comparison between Gd-BOPTA and Mangafodipir

The purpose of the study was to evaluate the MR contrast agents gadolinium benzyloxypropionictetro-acetate (Gd-BOPTA) and Mangafodipir for liver enhancement and the lesion-liver contrast on T1W spin-echo (SE) and gradient-recalled-echo (GRE) images. Fifty-one patients (three groups of 17 patients each) with known or suspected liver lesions were evaluated with T1W SE (300/12) and GRE (77-80/2.3-2.5/80 degrees) images before and after intravenous (IV) Gd-BOPTA (0.1 or 0.05 mmol/kg) or Mangafodipir (5 mumol/kg) in phase II to III clinical trials. Quantitative analysis by calculating liver signal-to-noise ratio (SNR), lesion-liver contrast-to-noise ratio (CNR), and spleen-liver CNR was performed. Liver SNR and spleen-liver CNR were always significantly increased postcontrast. SNR was highest after application of 0.1 mmol/kg Gd-BOPTA (51.3 +/- 3.6, P < .05). CNR was highest after Mangafodipir (-22.6 +/- 2.7), but this was not significantly different from others (P = .07). Overall, GRE images were superior to SE images for SNR and CNR. Mangafodipir and Gd-BOPTA (0.1 mmol/kg) provide equal liver enhancement and lesion conspicuity postcontrast. By all criteria, contrast-enhanced T1-weighted GRE were comparable to SE images.

Pulse sequence optimization for T2-weighted MR imaging of the brain

The authors implemented bipolar velocity compensated pulse techniques for T2-weighted MR imaging of the brain. Signal-to-noise (S/N) and image quality was compared for pulse sequences with standard and optimized RF pulses, low and regular bandwidth versions and cardiac triggering. Images from bipolar velocity compensated sequences allowed better visualization of vessels and basilar cisterns and improved image quality relative to standard sequences without velocity compensation. The implementation of optimized RF pulses with bipolar sequences resulted in further improvement in image quality. Single echo sequences consistently had improved image quality and signal-to-noise relative to the second echo of a double echo sequence. Low bandwidth bipolar sequences with extended sampling period had 30% higher S/N, but at the cost of slight loss in edge definition. The highest image quality was obtained with the bipolar, optimized RF, single echo sequence. Using this technique contiguous high quality image slices could be obtained with velocity compensation. The addition of cardiac triggering to bipolar sequences resulted in slight improvement in image quality, but this difference was marginal and probably rarely necessary for MR imaging of the brain.

Effect of the sampling rate on magnetic resonance imaging

Effects of the sampling rate on spatial resolution and the signal-to-noise ratio in spin-echo MR imaging were calculated. The theoretical results suggest that as T2* decreases, due to either static magnetic field inhomogeneities or shortened T2 relaxation times, the calculated optimum sampling rate increases accordingly. Since biological tissues exhibit wide variations in their magnetic susceptibilities and T2 relaxation times for various isotopes but modern magnets are characterized by good magnetic field homogeneity, the optimum sampling rate should exhibit an observable dependence on the tissue type. Proton and sodium imaging experiments confirm this conclusion.

Two-point T1 measurement: wide-coverage optimizations by stochastic simulations

Stochastic reliability of T1 measurement from image signal ratios is examined in the ideal case by stochastic simulations in the context of wide-coverage optimizations. Precise measurements prove to be accurate, and accurate ones precise. Sign-preserved inversion-recovery (IR)/non-IR techniques are the best ratio method, reciprocal non-IR/IR ones being equivalent, but inconvenient. Wide-coverage optima are relatively unsharp. Suggested guidelines for covering the 150- to 1500-ms T1 band are minimal relevant TE; TI about 400 ms; effective repetition times about in the ratio, TR2(IR)/TR1 (non-IR) = 2.5-3.0, and in a sum as long as possible up to about TR1 + TR2 = 3.5-4.0 s; signal-averaging after and only after TR1 + TR2 has been lengthened to the said region. Also suggested are different guidelines for covering T1 bands, 120-1200 and 200-1800 ms. Typically, precisions and accuracies improve linearly or faster with increasing S/N and (S/N)2, respectively. Unnecessarily high pixel resolutions or thin slicings exact great penalties in accuracies. Progressively shortening TR1 eventually transforms a wide coverage into a sharp targeting with small potential gains in a narrow T1 locality and large compromises almost everywhere else. The simulations yield an insight into applicabilities of standard error propagation analyses in two-point T1 measurement.

Dorsal midbrain syndrome in multiple sclerosis with magnetic resonance imaging correlation

We describe the clinical characteristics and a series of magnetic resonance imaging (MRI) studies in a patient with the features of dorsal midbrain syndrome occurring in the setting of multiple sclerosis. A T2-weighted MRI study revealed a discrete abnormality in the tectum of the midbrain whereas a high volume delayed computed tomography (CT) scan was uninformative. In parallel with remission of the clinical findings, the MRI abnormality diminished over time and was no longer visible at one year suggesting that some MRI detected MS lesions can completely disappear with time. This report demonstrates the use of MRI to detect and to follow sequentially sites of known disease activity in MS.

Magnetic resonance scanning in orbital tumor diagnosis

We have compared the accuracy of magnetic resonance scanning (MRI) versus computed tomography (CT) in the differentiation of lateral orbital masses. The MRI results did not improve our ability to accurately diagnose malignant epithelial and lymphoid tumors.

Optimizing tissue contrast in magnetic resonance imaging

Magnetic resonance imaging demands that tissue contrast and signal-to-noise advantages be sought in each component of the imaging system. One component of magnetic resonance imaging in which contrast and signal-to-noise ratios are easily manipulated is in the choice of pulse sequences and interpulse delay times. This article provides a general method for determining the best choices of interpulse delay times in pulse sequences and applies that method to saturation recovery, inversion recovery, and spin-echo sequences. Saturation recovery and inversion recovery sequences with rephasing pulses, and tissues with unequal hydrogen densities are considered. Optimization of pulse sequences is carried out for the two distinct cases of (a) a fixed number of sequence repetitions and (b) a fixed total imaging time. Analytic expressions are derived or approximate expressions are provided for the interpulse delay times that optimize contrast-to-noise ratios in each pulse sequence. The acceptable range of interpulse delay times to obtain reasonable contrast using each pulse sequence is discussed.

Selection of pulse sequences producing maximum tissue contrast in magnetic resonance imaging

The importance of spin density [N(H)] and spin-lattice (T1) and spin-spin (T2) relaxation in the characterization of tissue by nuclear magnetic resonance (NMR) is clearly recognized. This work considers which optimized pulse sequences provide the best tissue discrimination between a given pair of tissues. The effects of tissue spin density and machine-imposed minimum rephasing echo times (TEMIN) for achieving maximum signal tissue contrast are discussed. A long TEMIN sacrifices T1-dependent contrast in saturation recovery (SR) and inversion recovery (IR) pulse sequences so that spin-echo (SE) becomes the optimum sequence to provide tissue contrast, due to T2 relaxation. Pulse sequences providing superior performance may be selected based on spin density and T1 and T2 ratios for a given pair of tissues. Selection of the preferred pulse sequence and interpulse delay times to produce maximum tissue contrast is strongly dependent on knowledge of tissue spin densities as well as T1 and T2 characteristics. As the spin density ratio increases, IR replaces SR as the preferred sequence and SE replaces IR and SR as the pulse sequence providing superior contrast. To select the optimal pulse sequence and interpulse delay times, an accurate knowledge of tissue spin density, T1 and T2 must be known for each tissue.

Restoration of signal polarity in a set of inversion recovery NMR images

A method is described which enables unambiguous retrieval of sign information in a set of magnetic resonance magnitude images of the inversion recovery type. The proposed method starts from the observation that the inversion recovery curve S is a monotonically increasing function of the inversion time TI, and comes down to finding the zero-crossing time TI0 of this curve for each pixel within the image. Absolute Value(S)and S are then related by S(TI) = -Absolute Value(S(TI)) for TI =/< TI(0) and S(TI) = + Absolute Value(S(TI)) for TI =/> TI0. The method, which does not require additional knowledge with respect to any of the NMR parameters involved, is shown to be effective when at least four inversion recovery images with different inversion times of a particular object slice are available. The efficacy of sign retrieval is demonstrated by imaging experiments on phantoms and human subjects. The validity of the polarity restoration method is established by viewing its results against the results of conventional methods, i.e., NMR spectroscopy.