Pixel-by-pixel deconvolution of bolus-tracking data: optimization and implementation.

Quantification of haemodynamic parameters with a deconvolution analysis of bolus-tracking data is an ill-posed problem which requires regularization. In a previous study, simulated data without structural errors were used to validate two methods for a pixel-by-pixel analysis: standard-form Tikhonov regularization with either the L-curve criterion (LCC) or generalized cross validation (GCV) for selecting the regularization parameter. However, problems of image artefacts were reported when the methods were applied to patient data. The aim of this study was to investigate the nature of these problems in more detail and evaluate strategies of optimization for routine application in the clinic. In addition we investigated to which extent the calculation time of the algorithm can be minimized. In order to ensure that the conclusions are relevant for a larger range of clinical applications, we relied on patient data for evaluation of the algorithms. Simulated data were used to validate the conclusions in a more quantitative manner. We conclude that the reported problems with image quality can be removed by appropriate optimization of either LCC or GCV. In all examples this could be achieved with LCC without significant perturbation of the values in pixels where the regularization parameter was originally selected accurately. GCV could not be optimized for the renal data, and in the CT data only at the cost of image resolution. Using the implementations given, calculation times were sufficiently short for routine application in the clinic.

Deconvolution of bolus-tracking data: a comparison of discretization methods.

Model-free measurement of perfusion from bolus-tracking data requires a discretization of the tracer kinetic model. In this study a classification is provided of existing approaches to discretization, and the accuracy of these methods is compared. Two methods are included which are delay invariant (circulant and time shift) and three methods which are not (volterra, singular and hybrid). Simulations of magnetic resonance imaging (MRI) in the brain are performed for two tissue types (plug flow and compartment) with variable delay and dispersion times, temporal resolution and signal to noise. Simulations were compared to measurements in a patient data set. Both delay-invariant methods are equally accurate, but the circulant method is sensitive to data truncation. Overall volterra produces highest estimates of perfusion, followed by hybrid, singular and delay-invariant methods. Volterra is most accurate except in plug-flow without delay or dispersion, which represents an unrealistic tissue type. Differences between methods vanish when delay or dispersion times increase above the temporal resolution. It is concluded that when negative delays cannot be avoided or when an accurate estimate of left-right perfusion ratios is required, the time shift is the method of choice. When delays are certain to be positive and absolute accuracy is the objective, the volterra method is to be preferred.

Absolute quantification of 31P muscle metabolites using NMRS with an internal standard and a high-Q, double-tuned coil.

31P-nuclear magnetic resonance spectroscopy (NMRS) is used to determine absolute concentrations of 31P metabolites in rat muscle. A technique exploiting tissue water 1H as internal concentration reference and a highly sensitive, double-tuned coil is described and evaluated experimentally. On KH2PO4 solutions of known concentration and varying coil loading it is shown to allow neutralization of systematic errors due to conductive losses, which normally range up to 20% or more. In vivo application in the determination of the absolute concentrations of ATP, PCr and Pi in eight rat thighs yields results that are in good agreement with literature values.

Fat fraction of lumbar bone marrow using in vivo proton nuclear magnetic resonance spectroscopy.

Localized proton spectra of the human lumbar vertebral body were recorded in vivo at 1.5 T, using the STEAM (stimulated echo acquisition mode) pulse sequence. Thirty-seven patients (18 men, 19 women) were examined, ranging in age from 21 to 68 years (mean = 40, SD = 13). The fat fraction of the bone marrow was calculated from the areas of the fat and water peaks in the spectrum. The results demonstrate that, in the course of aging, there is a systematic increase in measured fat percentage of about 7% per decade of age. No sex difference could be established on the basis of our results.

31P-NMR spectroscopy and the metabolic properties of different muscle fibers.

To study the in vivo recruitment of different fiber types and their metabolic properties, 31P-nuclear magnetic resonance spectroscopy (31P-NMRS) of the human calf muscle was performed in seven normal sedentary subjects. In the exhaustive exercise protocol used, the work load was increased every minute during 5 min. This resulted in a prominent split of the Pi resonance in all subjects, indicating pH compartmentation in the muscles studied. From the chemical shift of the Pi peaks relative to phosphocreatine (PCr) at the end of the exercise, intracellular pH (pHi) averaged 6.92 +/- 0.05 (SD) in compartment 1 and 6.23 +/- 0.15 in compartment 2. The recovery of both Pi resonances after exercise could be followed easily in five of these subjects. The recovery rate of the Pi peak is a good estimate of the oxidative metabolism at the end of the exercise. A monoexponential regression analysis showed that the mean initial recovery rate S0 was 2.49 +/- 0.17%/s in compartment 1 and only 0.87 +/- 0.12%/s in compartment 2, indicating aerobic function three times higher in compartment 1 at the end of exercise. The mean relative ATP fraction dropped significantly (P less than 0.001), from 20.0 +/- 1.0% of the total 31P signal integral before exercise to 14.0 +/- 1.6% at the end of exercise. The simultaneous visualization of two compartments, in good order, one with high pHi and fast recovery and another with low pHi and slow recovery, is rationalized by the different metabolic behavior of type I and II fibers in human calf muscle in response to exhaustive exercise. This study demonstrates that 31P-NMRS is an excellent noninvasive procedure to quantify aerobic metabolism in both fiber types simultaneously.

A comparison between different imaging strategies for diffusion measurements with the centric phase-encoded turboFLASH sequence.

The study compared the results of three centrally reordered phase-encoded turboFLASH sequences for diffusion-weighted imaging (DWI). The sequences were conventional turboFLASH, turboFLASH with subtraction of T1-related effects, and turboFLASH with correction for T1-related effects during the imaging period only. The relative merits were studied with respect to image quality and accuracy by computer simulation and by experimental validation on phantoms and on in vivo rat brain. A T1-related underestimation of the diffusion coefficient ranging from -30% (T1 approximately 200 ms) to -5% (T1 approximately 1 s) was found to exist for the conventional sequence. Image artifacts, caused by longitudinal relaxation during the imaging period, are reflected in calculated diffusion maps. When the correction sequence is used, the artifacts and the systematic errors are reduced but longitudinal relaxation during the delay between preparation and imaging periods remains large enough to induce significant errors (-15% for T1 approximately 200 ms to -3% for T1 approximately 1 s). The subtraction sequence eliminates the influence of T1 effects on the calibrations, but leads to identical artifacts for all diffusion-weighted images.

Optimization of sequence parameters in fast MR imaging of the brain with FLASH.

Owing to the intrinsically complex behavior of the signal intensity of fast gradient-refocusing MR sequences, agreement as to the clinically most useful sequence parameters has not yet been reached. This study evaluates the FLASH (fast low-angle shot) sequence for gray-white matter differentiation on normal volunteers at 1.5 T. The FLASH gradient-echo sequence is essentially T1-dependent. For very fast imaging and T1 weighting, the following parameters yield the best results: a flip angle of 30-50 degrees with TR = 20 and TE = 10. To replace T1-weighted SE by the faster FLASH sequence, the best results are achieved by a flip angle of 70-120 degrees with TR = 150-300 and TE = 10 (or shorter, if possible). The most valuable proton-density aspect is achieved by a flip angle of 30 degrees with TR = 300 and TE = 16.

Diffusion-weighted MR imaging of intracerebral masses: comparison with conventional MR imaging and histologic findings.

BACKGROUND AND PURPOSE: The purposes of this study were to find the role of diffusion-weighted MR imaging in characterizing intracerebral masses and to find a correlation, if any, between the different parameters of diffusion-weighted imaging and histologic analysis of tumors. The usefulness of diffusion-weighted imaging and apparent diffusion coefficient (ADC) maps in tumor delineation was evaluated. Contrast with white matter and ADC values for tumor components with available histology were also evaluated. METHODS: Twenty patients with clinical and routine MR imaging/CT evidence of intracerebral neoplasm were examined with routine MR imaging and echo-planar diffusion-weighted imaging. The routine MR imaging included at least the axial T2-weighted fast spin-echo and axial T1-weighted spin-echo sequences before and after contrast enhancement. The diffusion-weighted imaging included an echo-planar spin-echo sequence with three b values (0, 300, and 1200 s/mm(2)), sensitizing gradient in the z direction, and calculated ADC maps. The visual comparison of routine MR images with diffusion-weighted images for tumor delineation was performed as was the statistical analysis of quantitative diffusion-weighted imaging parameters with histologic evaluation. RESULTS: For tumors, the diffusion-weighted images and ADC maps of gliomas were less useful than the T2-weighted spin-echo and contrast-enhanced T1-weighted spin-echo images in definition of tumor boundaries. Additionally, in six cases of gliomas, neither T2-weighted spin-echo nor diffusion-weighted images were able to show a boundary between tumor and edema, which was present on contrast-enhanced T1-weighted and/or perfusion echo-planar images. The ADC values of solid gliomas, metastases, and meningioma were in the same range. In two cases of lymphomas, there was a good contrast with white matter, with strongly reduced ADC values. For infection, the highest contrast on diffusion-weighted images and lowest ADC values were observed in association with inflammatory granuloma and abscess. CONCLUSION: Contrary to the findings of previous studies, we found no clear advantage of diffusion-weighted echo-planar imaging in the evaluation of tumor extension. The contrast between gliomas, metastases, meningioma, and white matter was generally lower on diffusion-weighted images and ADC maps compared with conventional MR imaging. Unlike gliomas, the two cases of lymphomas showed hyperintense signal on diffusion-weighted images whereas the case of cerebral abscess showed the highest contrast on diffusion-weighted images with very low ADC values. Further study is required to find out whether this may be useful in the differentiation of gliomas and metastasis from lymphoma and abscess.

Choice of the regularization parameter for perfusion quantification with MRI.

Truncated singular value decomposition (TSVD) is an effective method for the deconvolution of dynamic contrast enhanced (DCE) MRI. Two robust methods for the selection of the truncation threshold on a pixel-by-pixel basis--generalized cross validation (GCV) and the L-curve criterion (LCC)--were optimized and compared to paradigms in the literature. GCV and LCC were found to perform optimally when applied with a smooth version of TSVD, known as standard form Tikhonov regularization (SFTR). The methods lead to improvements in the estimate of the residue function and of its maximum, and converge properly with SNR. The oscillations typically observed in the solution vanish entirely, and perfusion is more accurately estimated at small mean transit times. This results in improved image contrast and increased sensitivity to perfusion abnormalities, at the cost of 1-2 min in calculation time and hyperintense clusters in the image. Preliminary experience with clinical data suggests that the latter problem can be resolved using spatial continuity and/or hybrid thresholding methods. In the simulations GCV and LCC are equivalent in terms of performance, but GCV thresholding is faster.

Quantification of trabecular structure in the distal femur using magnetic resonance phase imaging.

A new approach for quantifying trabecular bone tissue using the phase images of a simple gradient-echo sequence is presented. The proposed method is based on the hypothesis that the differences in susceptibility between bone and bone marrow cause magnetic field (i.e., precession phase) variations between the image voxels. Phase images of the distal femur were obtained in vivo and characterised with the use of the phase variance. Computer simulations and experimental results indicate that the distribution of the phases varies with echo time and image resolution, as expected. Keeping these fixed, however, the phase variance is found to strongly reflect variations in trabecular structure.

A method for myelin fiber orientation mapping using diffusion-weighted MR images.

In the past, the anisotropic diffusion of water molecules in white matter in the brain has been correlated to the basic symmetry of the myelin fibers: water diffuses more readily along the fiber direction than perpendicular to it. As a consequence, diffusion sensitized magnetic resonance imaging can be expected to be useful for studying the fiber orientation. In this work, we present a method for exploiting this type of information to map the fiber orientations in the image plane. It makes use of three diffusion-weighted images with sensitizing gradients along x, y and u, an axis at 45 degrees with respect to x and y. The orientation information contained in these images is summarized in a single image representing the angle between the fiber direction and a fixed axis, making use of a cyclic color scale. The method is evaluated using computer simulations for a variety of diffusion weighting strengths and signal-to-noise ratios, tested on a phantom and illustrated on an in vivo example. An extension to the determination of the fiber orientation in three dimensions is also described.

Comparison of computer simulated and phantom measured phase variance in the study of trabecular bone.

A new method using magnetic resonance phase images for the assessment of trabecular bone structure has recently been proposed. To test this method, a mathematical model is developed which calculates the phase distribution in gradient echo acquired phase images of a structure of Pyrex glass rods immersed in a copper sulfate solution. Several experiments were performed using a phantom built in the same way as the structure used in the mathematical model. The results from the model are compared with those from the phantom tests, and the influence of resolution and bone area fraction on the phase dispersion is studied. The good correlation between theoretical and experimental results shows that phase variance increases with increasing resolution and bone density. However, the dependence of variance on bone density is less prominent for large pixel sizes.

MR phase imaging to quantify bone volume fraction: computer simulations and in vivo measurements.

Magnetic resonance phase images can be used to assess trabecular bone by measuring the standard deviation of the phases in a region of interest. The standard deviation of regional phase measurements reflects the degree of magnetic field inhomogeneity caused by susceptibility differences between bone and marrow. A 3D computer model of trabecular bone was developed and then used to explore the influence of bone volume fraction and imaging parameters such as pixel size and slice width on the standard deviation of regional phase measurements. The results from these tests show that with appropriate selection of these parameters, phase spread strongly reflects variations in trabecular bone density (a correlation of R(2) = 0.98 with bone volume fraction between 0 and 10%). The technique was then applied in vivo on the radius of 25 patients who already had a bone density scan with peripheral quantitative tomography and a correlation between phase standard deviation and trabecular bone density was found (R(2) = 0.46).

Multicentre magnetic resonance texture analysis trial using reticulated foam test objects.

Texture analysis in magnetic resonance imaging has the ability to provide useful diagnostic information with respect to the discrimination of disease states of a single tissue or the separation of different tissues. However, for widespread use it is necessary to determine how texture measurements carried out in one center relate to those carried out in another. To this end, a multicentre trial has been performed where reticulated foam test objects have been scanned in six European centers according to a fixed protocol. It has been concluded that texture measurements are not transportable between centers. Principal component models calculated from the texture parameters collected in one center do not fit the data collected in another. Further trials are to investigate whether the reticulated foam test objects may be used to normalize tissue texture data collected in different centers.

Reproducibility of CT bone densitometry: operator versus automated ROI definition.

Intrasubject reproducibility with repeated determination of vertebral mineral density from a given set of CT images was investigated. The region of interest (ROI) in 10 patient scans was selected by four independent operators either manually or with an automated procedure separating cortical and spongeous bone, the operators being requested to interact in ROI selection. The mean intrasubject variation was found to be much lower with the automated process (0.3 to 0.6%) than with the conventional method (2.5 to 5.2%). In a second study, 10 patients were examined twice to determine the reproducibility of CT slice selection by the operator. The errors were of the same order of magnitude as in ROI selection.

Diffusion and perfusion MRI: basic physics.

Diffusion and perfusion MR imaging are now being used increasingly in neuro-vascular clinical applications. While diffusion weighted magnetic resonance imaging exploits the translational mobility of water molecules to obtain information on the microscopic behaviour of the tissues (presence of macromolecules, presence and permeability of membranes, equilibrium intracellular-extracellular water, ellipsis), perfusion weighted imaging makes use of endogenous and exogenous tracers for monitoring their hemodynamic status. The combination of both techniques is extremely promising for the early detection and assessment of stroke, for tumor characterisation and for the evaluation of neurodegenerative diseases. This article provides a brief review of the basic physics principles underlying the methodologies followed.

Benign liver lesions: differentiation by magnetic resonance.

Optimal hepatic imaging involves both detection and characterization of focal lesions. Detection involves both determination of the presence of lesions and of their segmental extent of liver involvement. In the evaluation of hypervascular lesions, magnetic resonance imaging (MRI) has a greater impact on patient management than ultrasound (US) and computed tomography (CT). Most benign tumors are incidental findings and do not produce clinical symptoms. They must be accurately diagnosed without using aggressive procedures. Knowledge of their imaging features is essential to avoid unnecessary work-up and to minimize patient anxiety. In this article, the MR appearance, vascular and functional behavior of the most common benign liver tumors will be discussed.

CT-assisted chemonucleolysis.

The safety and success rate of chemonucleolysis are largely dependent on correct needle insertion. The neurological complications of chemonucleolysis are related to inadvertent puncture of the dura, nerve root and/or spinal nerve. To avoid this complication, a puncture angle of 50 degrees to 60 degrees above the sagittal plane is widely recommended (1). Our CT study of the location of the spinal nerve L4-L5 and L5-S1 demonstrates a great variability of this location and also explains some of the possible hazards of 50 degrees to 60 degrees puncture (2). The CT-evaluation of the relationship between the nerve root, spinal nerve and apophyseal joint with respect to the puncture angle is suitable. In our experience, this method is helpful for fast, safe and easy insertion of the needle.

Pituitary microadenomas: diagnosis with dynamic serial CT, conventional CT and T1-weighted MR imaging before and after injection of gadolinium.

PURPOSE: The few studies comparing CT and MR in the diagnosis of pituitary microadenomas give discordant results. This prospective study compares the value of dynamic serial CT (DSCT), conventional CT (CCT), pre-Gd T1-weighted spin echo (T1SE-Gd) and post-Gd T1-weighted SE (T1SE+Gd) in the detection of pituitary microadenomas. METHODS: The value of CT and MR was assessed in 17 cases, with the surgical confirmation as reference. The CTs and MRs were independently and blindly evaluated by two neuroradiologists. Because the finding at MR and/or CT influenced the selection of patients for surgery, the positive predictive value was calculated. The sensitivity calculation applies only for a specific group of postsurgical patients with positive CT and/or MR (marked with an asterisk for these reasons). RESULTS: In this study, the highest sensitivity* (88% and 82%, respectively, for observers I and II) was achieved by DSCT. The T1SE-Gd and T1SE+Gd (evaluated separately) were slightly inferior (sensitivity* of 76% for observer I). The combination of T1SE-Gd with T1SE+Gd improved the sensitivity* to 94% and 82% respectively, for observers I and II. There was also moderate improvement of sensitivity* by combination of DSCT with conventional CT (94% for observer I). CONCLUSION: These results suggest that rapid injection of a large amount of contrast conjointly with a fast scanning through the pituitary is crucial for the confident detection of microprolactinomas. Taking into account the similar results of the combination DSCT + CT and T1SE-Gd + T1SE+Gd in the detection of microprolactinomas, high field MRI is the first step modality in the evaluation of presumed hypophyseal hyperprolactinemia.

Quantitative ultrasound of the calcaneus: an in vivo comparison with dual-energy X-ray absorptiometry and magnetic resonance imaging.

The current study was performed in a clinical setting and aimed to evaluate the relationship between quantitative ultrasound (QUS) of the calcaneus with bone mineral density (BMD) assessed with dual-energy X-ray absorptiometry (DXA) and with variables derived from magnetic resonance imaging (MRI). Thirty-two postmenopausal women (mean age 61 years) were studied at the level of the nondominant calcaneus. QUS was performed using a DTU-one device including parametric imaging and yielded speed of sound (SOS) and broadband ultrasound attenuation (BUA) data. DXA was performed at a matched region of interest (ROI) in the calcaneus, using a Hologic QDR 4500 device. MRI, also performed at a matched ROI, yielded, using a Siemens Magnetom Vision device, the inverse of the transverse relaxation time (1/T(2)(*)) and the phase standard deviation (PSD). The strongest relationship between QUS and the other variables involved BUA and BMD (r &equals: 0.677, p < 0.001); 1/T(2)(*) showed a trend to correlation with SOS (r = 0.359, p = 0. 044) and with BMD (r = 0.364, p = 0.040), while the relationship between 1/T(2)(*) and BUA, PSD and BUA, PSD and SOS, PSD and BMD remained far from significance. Regression analysis of QUS, DXA, and MRI variables against age showed a trend to significant decline only for 1/T(2)(*) (r = -0.409, p = 0.020). In conclusion, this study shows that BUA of the calcaneus has the best correlation with BMD, and that, at least in a clinical setting, the ability of QUS to give information about bone structure is limited.