Blind deconvolution decreases requirements on temporal resolution of DCE-MRI: Application to 2nd generation pharmacokinetic modeling

PURPOSE

Dynamic Contrast-Enhanced (DCE) MRI with 2nd generation pharmacokinetic models provides estimates of plasma flow and permeability surface-area product in contrast to the broadly used 1st generation models (e.g. the Tofts models). However, the use of 2nd generation models requires higher frequency with which the dynamic images are acquired (around 1.5 s per image). Blind deconvolution can decrease the demands on temporal resolution as shown previously for one of the 1st generation models. Here, the temporal-resolution requirements achievable for blind deconvolution with a 2nd generation model are studied.

METHODS

The 2nd generation model is formulated as the distributed-capillary adiabatic-tissue-homogeneity (DCATH) model. Blind deconvolution is based on Parker's model of the arterial input function. The accuracy and precision of the estimated arterial input functions and the perfusion parameters is evaluated on synthetic and real clinical datasets with different levels of the temporal resolution.

RESULTS

The estimated arterial input functions remained unchanged from their reference high-temporal-resolution estimates (obtained with the sampling interval around 1 s) when increasing the sampling interval up to about 5 s for synthetic data and up to 3.6-4.8 s for real data. Further increasing of the sampling intervals led to systematic distortions, such as lowering and broadening of the 1st pass peak. The resulting perfusion-parameter estimation error was below 10% for the sampling intervals up to 3 s (synthetic data), in line with the real data perfusion-parameter boxplots which remained unchanged up to the sampling interval 3.6 s.

CONCLUSION

We show that use of blind deconvolution decreases the demands on temporal resolution in DCE-MRI from about 1.5 s (in case of measured arterial input functions) to 3-4 s. This can be exploited in increased spatial resolution or larger organ coverage.

T1 mapping of myocardium in rats using self-gated golden-angle acquisition

PURPOSE

The aim of this study is to design a method of myocardial T1 quantification in small laboratory animals and to investigate the effects of spatiotemporal regularization and the needed acquisition duration.

METHODS

We propose a compressed-sensing approach to T1 quantification based on self-gated inversion-recovery radial two/three-dimensional (2D/3D) golden-angle stack-of-stars acquisition with image reconstruction performed using total-variation spatiotemporal regularization. The method was tested on a phantom and on a healthy rat, as well as on rats in a small myocardium-remodeling study.

RESULTS

The results showed a good match of the T1 estimates with the results obtained using the ground-truth method on a phantom and with the literature values for rats myocardium. The proposed 2D and 3D methods showed significant differences between normal and remodeling myocardium groups for acquisition lengths down to approximately 5 and 15 min, respectively.

CONCLUSIONS

A new 2D and 3D method for quantification of myocardial T1 in rats was proposed. We have shown the capability of both techniques to distinguish between normal and remodeling myocardial tissue. We have shown the effects of image-reconstruction regularization weights and acquisition length on the T1 estimates.

Feasibility of Hepatic Fat Quantification Using Proton Density Fat Fraction by Multi-Echo Chemical-Shift-Encoded MRI at 7T

Fat fraction quantification and assessment of its distribution in the hepatic tissue become more important with the growing epidemic of obesity, and the increasing prevalence of diabetes mellitus type 2 and non-alcoholic fatty liver disease. At 3Tesla, the multi-echo, chemical-shift-encoded magnetic resonance imaging (CSE-MRI)-based acquisition allows the measurement of proton density fat-fraction (PDFF) even in clinical protocols. Further improvements in SNR can be achieved by the use of phased array coils and increased static magnetic field. The purpose of the study is to evaluate the feasibility of PDFF imaging using a multi-echo CSE-MRI technique at ultra-high magnetic field (7Tesla). Thirteen volunteers (M/F) with a broad range of age, body mass index, and hepatic PDFF were measured at 3 and 7T by multi-gradient-echo MRI and single-voxel spectroscopy MRS. All measurements were performed in breath-hold (exhalation); the MRI protocols were optimized for a short measurement time, thus minimizing motion-related problems. 7T data were processed off-line using Matlab® (MRI:multi-gradient-echo) and jMRUI (MRS), respectively. For quantitative validation of the PDFF results, a similar protocol was performed at 3T, including on-line data processing provided by the system manufacturer, and correlation analyses between 7 and 3T data were performed off-line. The multi-echo CSE-MRI measurements at 7T with a phased-array coil configuration and an optimal post-processing yielded liver volume coverage ranging from 30 to 90% for high- and low-BMI subjects, respectively. PDFFs ranged between 1 and 20%. We found significant correlations between 7T MRI and -MRS measurements (R² ≅ 0.97; p < 0.005), and between MRI-PDFF at 7T and 3T fields (R² ≅ 0.94; p < 0.005) in the evaluated volumes. Based on the measurements and analyses performed, the multi-echo CSE-MRI method using a 32-channel coil at 7T showed its aptitude for MRI-based quantitation of PDFF in the investigated volumes. The results are the first step toward qMRI of the whole liver at 7T with further improvements in hardware.

Contribution of macromolecules to brain 1 H MR spectra: Experts' consensus recommendations

Proton MR spectra of the brain, especially those measured at short and intermediate echo times, contain signals from mobile macromolecules (MM). A description of the main MM is provided in this consensus paper. These broad peaks of MM underlie the narrower peaks of metabolites and often complicate their quantification but they also may have potential importance as biomarkers in specific diseases. Thus, separation of broad MM signals from low molecular weight metabolites enables accurate determination of metabolite concentrations and is of primary interest in many studies. Other studies attempt to understand the origin of the MM spectrum, to decompose it into individual spectral regions or peaks and to use the components of the MM spectrum as markers of various physiological or pathological conditions in biomedical research or clinical practice. The aim of this consensus paper is to provide an overview and some recommendations on how to handle the MM signals in different types of studies together with a list of open issues in the field, which are all summarized at the end of the paper.

Simple compensation method for improved half-pulse excitation profile with rephasing gradient

PURPOSE

To improve the slice profile quality obtained by RF half-pulse excitation for 2D-UTE applications.

METHODS

The overall first-order and zero-order phase errors along the slice-selection direction were obtained with the help of an optimization task to minimize the out-of-slice signal contamination from the calibration 1-dimenisonal (1D) profile data. The time-phase-error evolution was approximated from the k-space readout data, which were acquired primarily for correction of the readout trajectories during data regridding to the rectilinear grids. The correction of the slice profile was achieved by rephasing gradient pulses applied immediately after the end of excitation. The total prescan calibration typically took less than 2 minutes.

RESULTS

The improved image quality using the proposed calibration method was demonstrated both on phantoms and on ankle images obtained from healthy volunteers. It was demonstrated that calibration can be performed either as a separate water phantom measurement or directly as a prescan procedure.

CONCLUSION

The slice-profile distortion from the half-pulse excitation could substantially affect the overall fidelity of 2D-UTE images. The presented experiments proved that the image quality could be substantially increased by application of the proposed slice-correction method.

Low-level fat fraction quantification at 3 T: comparative study of different tools for water-fat reconstruction and MR spectroscopy

OBJECTIVES

Chemical Shift Encoded Magnetic Resonance Imaging (CSE-MRI)-based quantification of low-level (< 5% of proton density fat fraction-PDFF) fat infiltration requires highly accurate data reconstruction for the assessment of hepatic or pancreatic fat accumulation in diagnostics and biomedical research.

MATERIALS AND METHODS

We compare three software tools available for water/fat image reconstruction and PDFF quantification with MRS as the reference method. Based on the algorithm exploited in the tested software, the accuracy of fat fraction quantification varies. We evaluate them in phantom and in vivo MRS and MRI measurements.

RESULTS

The signal model of Intralipid 20% emulsion used for phantoms was established for 3 T and 9.4 T fields. In all cases, we noticed a high coefficient of determination (R-squared) between MRS and MRI-PDFF measurements: in phantoms <0.9924-0.9990>; and in vivo <0.8069-0.9552>. Bland-Altman analysis was applied to phantom and in vivo measurements.

DISCUSSION

Multi-echo MRI in combination with an advanced algorithm including multi-peak spectrum modeling appears as a valuable and accurate method for low-level PDFF quantification over large FOV in high resolution, and is much faster than MRS methods. The graph-cut algorithm (GC) showed the fewest water/fat swaps in the PDFF maps, and hence stands out as the most robust method of those tested.

Influence of k-space trajectory corrections on proton density mapping with ultrashort echo time imaging: Application for imaging of short T2 components in white matter

PURPOSE

To evaluate the impact of MR gradient system imperfections and limitations for the quantitative mapping of short T2* signals performed by ultrashort echo time (UTE) acquisition approach.

MATERIALS AND METHODS

The measurement of short T2* signals from a phantom and a healthy volunteer study (8 subjects of average age 28 ± 4 years) were performed on a 3T scanner. The characteristics of the gradient system were obtained using calibration method performed directly on the measured subject or phantom. This information was used to calculate the actual sampling trajectory with the help of a parametric eddy current model. The actual sample positions were used to reconstruct corrected images and compared with uncorrected data.

RESULTS

Comparison of both approaches, i.e., without and with correction of k-space sampling trajectories revealed substantial improvement when correction was applied. The phantom experiments demonstrate substantial in-plane signal intensity variations for uncorrected sampling trajectories. In the case of the volunteer study, this led to significant differences in relative proton density (RPD) estimation between the uncorrected and corrected data (P = 0.0117 by Wilcoxon matched-pairs test) and provides for about ~15% higher values for short T2* components of white matter (WM) in the case of uncorrected images.

CONCLUSION

The imperfection of the applied gradients could induce errors in k-space data sampling which further propagates into the fidelity of the UTE images and jeopardizes precision of quantification. However, the study proved that measurement of gradient errors together with correction of sample positions can contribute to increased accuracy and unbiased characterization of short T2* signals.

Processing tracking in jMRUI software for magnetic resonance spectra quantitation reproducibility assurance

Background

Proton magnetic resonance spectroscopy is a non-invasive measurement technique which provides information about concentrations of up to 20 metabolites participating in intracellular biochemical processes. In order to obtain any metabolic information from measured spectra a processing should be done in specialized software, like jMRUI. The processing is interactive and complex and often requires many trials before obtaining a correct result. This paper proposes a jMRUI enhancement for efficient and unambiguous history tracking and file identification.

Results

A database storing all processing steps, parameters and files used in processing was developed for jMRUI. The solution was developed in Java, authors used a SQL database for robust storage of parameters and SHA-256 hash code for unambiguous file identification. The developed system was integrated directly in jMRUI and it will be publically available. A graphical user interface was implemented in order to make the user experience more comfortable. The database operation is invisible from the point of view of the common user, all tracking operations are performed in the background.

Conclusions

The implemented jMRUI database is a tool that can significantly help the user to track the processing history performed on data in jMRUI. The created tool is oriented to be user-friendly, robust and easy to use. The database GUI allows the user to browse the whole processing history of a selected file and learn e.g. what processing lead to the results, where the original data are stored, to obtain the list of all processing actions performed on spectra.

Distributed capillary adiabatic tissue homogeneity model in parametric multi-channel blind AIF estimation using DCE-MRI

PURPOSE

One of the main challenges in quantitative dynamic contrast-enhanced (DCE) MRI is estimation of the arterial input function (AIF). Usually, the signal from a single artery (ignoring contrast dispersion, partial volume effects and flow artifacts) or a population average of such signals (also ignoring variability between patients) is used.

METHODS

Multi-channel blind deconvolution is an alternative approach avoiding most of these problems. The AIF is estimated directly from the measured tracer concentration curves in several tissues. This contribution extends the published methods of multi-channel blind deconvolution by applying a more realistic model of the impulse residue function, the distributed capillary adiabatic tissue homogeneity model (DCATH). In addition, an alternative AIF model is used and several AIF-scaling methods are tested.

RESULTS

The proposed method is evaluated on synthetic data with respect to the number of tissue regions and to the signal-to-noise ratio. Evaluation on clinical data (renal cell carcinoma patients before and after the beginning of the treatment) gave consistent results. An initial evaluation on clinical data indicates more reliable and less noise sensitive perfusion parameter estimates.

CONCLUSION

Blind multi-channel deconvolution using the DCATH model might be a method of choice for AIF estimation in a clinical setup.

Blind deconvolution in dynamic contrast-enhanced MRI and ultrasound

This paper is focused on quantitative perfusion analysis using MRI and ultrasound. In both MRI and ultrasound, most approaches allow estimation of rate constants (Ktrans, kep for MRI) and indices (AUC, TTP) that are only related to the physiological perfusion parameters of a tissue (e.g. blood flow, vessel permeability) but do not allow their absolute quantification. Recent methods for quantification of these physiological perfusion parameters are shortly reviewed. The main problem of these methods is estimation of the arterial input function (AIF). This paper summarizes and extends the current blind-deconvolution approaches to AIF estimation. The feasibility of these methods is shown on a small preclinical study using both MRI and ultrasound.