New insight into the contrast in diffusional kurtosis images: Does it depend on magnetic susceptibility?

Influence of adaptive denoising on Diffusion Kurtosis Imaging at 3T and 7T

BACKGROUND AND OBJECTIVE

Choosing the most appropriate denoising method to improve the quality of diagnostic images maximally is key in pre-processing of diffusion MRI images. Recent advancements in acquisition and reconstruction techniques have questioned traditional noise estimation methods favoring adaptive denoising frameworks, circumventing the need to know a priori information that is hardly available in a clinical setting. In this observational study, we compared two innovative adaptive techniques sharing some features, Patch2Self and Nlsam, through application on reference adult data at 3T and 7T. The primary aim was identifying the most effective method in case of Diffusion Kurtosis Imaging (DKI) data - particularly susceptible to noise and signal fluctuations - at 3T and 7T fields. A side goal consisted of investigating the dependence of kurtosis metrics' variability with respect to the magnetic field on the adopted denoising methodology.

METHODS

For comparison purposes, we focused on qualitative and quantitative analysis of DKI data and related microstructural maps before and after applying the two denoising approaches. Specifically, we assessed computational efficiency, preservation of anatomical details via perceptual metrics, consistency of microstructure model fitting, alleviation of degeneracies in model estimation, and joint variability with varying field strength and denoising method.

RESULTS

Accounting for all these factors, Patch2Self framework has turned out to be specifically suitable for DKI data, with improving performance at 7T. Nlsam method is more robust in alleviating degenerate black voxels while introducing some blurring, which in turn is reflected in an overall loss of image sharpness. Regarding the impact of denoising on field-dependent variability, both methods have been shown to make variations from standard to Ultra-High Field more concordant with theoretical evidence, claiming that kurtosis metrics are sensitive to susceptibility-induced background gradients, directly proportional to the magnetic field strength and sensitive to the microscopic distribution of iron and myelin.

CONCLUSIONS

This study serves as a proof-of-concept stressing the need for an accurate choice of a denoising methodology, specifically tailored for the data under analysis and allowing higher spatial resolution acquisition within clinically compatible timings, with all the potential benefits that improving suboptimal quality of diagnostic images entails.

Nuclear magnetic resonance signal decay in the presence of a background gradient: Normal and anomalous diffusion

A novel way for calculating the diffusion-weighted nuclear magnetic resonance (NMR) attenuation signal expression in the presence of a background gradient is developed. This method is easily applicable to NMR-attenuated signals arising from any pulse field gradient sequence experiments. Here, we provide detailed calculations for the classical pulsed gradient stimulated echo and the pulsed gradient spin echo, as the particular cases. Within this general theoretical framework, devised for Gaussian processes with stationary increments, we recover and extend the previous Stejskal-Tanner results in the case of normal diffusion and we furnish a new expression in the case of anomalous diffusion.

Feasibility of in vivo multi-parametric quantitative magnetic resonance imaging of the healthy sciatic nerve with a unified signal readout protocol

Magnetic resonance neurography (MRN) has been used successfully over the years to investigate the peripheral nervous system (PNS) because it allows early detection and precise localisation of neural tissue damage. However, studies demonstrating the feasibility of combining MRN with multi-parametric quantitative magnetic resonance imaging (qMRI) methods, which provide more specific information related to nerve tissue composition and microstructural organisation, can be invaluable. The translation of emerging qMRI methods previously validated in the central nervous system to the PNS offers real potential to characterise in patients in vivo the underlying pathophysiological mechanisms involved in a plethora of conditions of the PNS. The aim of this study was to assess the feasibility of combining MRN with qMRI to measure diffusion, magnetisation transfer and relaxation properties of the healthy sciatic nerve in vivo using a unified signal readout protocol. The reproducibility of the multi-parametric qMRI protocol as well as normative qMRI measures in the healthy sciatic nerve are reported. The findings presented herein pave the way to the practical implementation of joint MRN-qMRI in future studies of pathological conditions affecting the PNS.

Developing a Predictive Grading Model for Children with Gliomas Based on Diffusion Kurtosis Imaging Metrics: Accuracy and Clinical Correlations with Patient Survival

Purpose: To develop a predictive grading model based on diffusion kurtosis imaging (DKI) metrics in children affected by gliomas, and to investigate the clinical impact of the predictive model by correlating with overall survival and progression-free survival. Materials and methods: 59 patients with a histological diagnosis of glioma were retrospectively studied (33 M, 26 F, median age 7.2 years). Patients were studied on a 3T scanner with a standardized MR protocol, including conventional and DKI sequences. Mean kurtosis (MK), axial kurtosis (AK), radial kurtosis (RK), fractional anisotropy (FA), and apparent diffusion coefficient (ADC) maps were obtained. Whole tumour volumes (VOIs) were segmented semi-automatically. Mean DKI values were calculated for each metric. The quantitative values from DKI-derived metrics were used to develop a predictive grading model to develop a probability prediction of a high-grade glioma (pHGG). Three models were tested: DTI-based, DKI-based, and combined (DTI and DKI). The grading accuracy of the resulting probabilities was tested with a receiver operating characteristics (ROC) analysis for each model. In order to account for dataset imbalances between pLGG and pHGG, we applied a random synthetic minority oversampling technique (SMOTE) analysis. Lastly, the most accurate model predictions were correlated with progression-free survival (PFS) and overall survival (OS) using the Kaplan−Meier method. Results: The cohort included 46 patients with pLGG and 13 patients with pHGG. The developed model predictions yielded an AUC of 0.859 (95%CI: 0.752−0.966) for the DTI model, of 0.939 (95%CI: 0.879−1) for the DKI model, and of 0.946 (95%CI: 0.890−1) for the combined model, including input from both DTI and DKI metrics, which resulted in the most accurate model. Sample estimation with the random SMOTE analysis yielded an AUC of 0.98 on the testing set. Model predictions from the combined model were significantly correlated with PFS (25.2 months for pHGG vs. 40.0 months for pLGG, p < 0.001) and OS (28.9 months for pHGG vs. 44.9 months for pLGG, p < 0.001). Conclusions: a DKI-based predictive model was highly accurate for pediatric glioma grading. The combined model, derived from both DTI and DKI metrics, proved that DKI-based model predictions of tumour grade were significantly correlated with progression-free survival and overall survival.

Performance of Diffusion Kurtosis Imaging Versus Diffusion Tensor Imaging in Discriminating Between Benign Tissue, Low and High Gleason Grade Prostate Cancer

RATIONALE AND OBJECTIVES

To investigate the performance of diffusion kurtosis imaging (DKI) and diffusion tensor imaging (DTI) in discriminating benign tissue, low- and high-grade prostate adenocarcinoma (PCa).

MATERIALS AND METHODS

Forty-eight patients with biopsy-proven PCa of different Gleason grade (GG), who provided written informed consent, were enrolled. All subjects underwent 3T DWI examinations by using b values 0, 500, 1000, 1500, 2000, and 2500 s/mm² and six gradient directions. Mean diffusivity, fractional anisotropy (FA), apparent kurtosis (K), apparent kurtosis-derived diffusivity (D), and proxy fractional kurtosis anisotropy (KFA) maps were obtained. Regions of interest were selected in PCa, in the contralateral benign zone, and in the peritumoral area. Histogram analysis was performed by measuring mean, 10th, 25th, and 90th (p90) percentile of the whole-lesion volume. Kruskal-Wallis test with Bonferroni correction was used to assess significant differences between different regions of interest. The correlation between diffusion metrics and GG and between DKI and DTI parameters was evaluated with Pearson's test. ROC curve analysis was carried out to analyze the ability of histogram variables to differentiate low- and high-GG PCa.

RESULTS

All metrics significantly discriminated PCa from benign and from peritumoral tissue (except for K, KFA_p90, and FA). K_p90 showed the highest correlation with GG and the best diagnostic ability (area under the curve = 0.84) in discriminating low- from high-risk PCa.

CONCLUSION

Compared to DTI, DKI provides complementary and additional information about prostate cancer tissue, resulting more sensitive to PCa-derived modifications and more accurate in discriminating low- and high-risk PCa.

Age-related microstructural and physiological changes in normal brain measured by MRI γ-metrics derived from anomalous diffusion signal representation

Nowadays, increasing longevity associated with declining cerebral nervous system functions, suggests the need for continued development of new imaging contrast mechanisms to support the differential diagnosis of age-related decline. In our previous papers, we developed a new imaging contrast metrics derived from anomalous diffusion signal representation and obtained from diffusion-weighted (DW) data collected by varying diffusion gradient strengths. Recently, we highlighted that the new metrics, named γ-metrics, depended on the local inhomogeneity due to differences in magnetic susceptibility between tissues and diffusion compartments in young healthy subjects, thus providing information about myelin orientation and iron content within cerebral regions. The major structural modifications occurring in brain aging are myelinated fibers damage in nerve fibers and iron accumulation in gray matter nuclei. Therefore, we investigated the potential of γ-metrics in relation to other conventional diffusion metrics such as DTI, DKI and NODDI in detecting age-related structural changes in white matter (WM) and subcortical gray matter (scGM). DW-images were acquired in 32 healthy subjects, adults and elderly (age range 20-77 years) using 3.0T and 12 b-values up to 5000 s/mm². Association between diffusion metrics and subjects' age was assessed using linear regression. A decline in mean γ (Mγ) in the scGM and a complementary increase in radial γ (γ_⊥) in frontal WM, genu of corpus callosum and anterior corona radiata with advancing age were found. We suggested that the increase in γ⊥ might reflect declined myelin density, and Mγ decrease might mirror iron accumulation. An increase in D_// and a decrease in the orientation dispersion index (ODI) were associated with axonal loss in the pyramidal tracts, while their inverted trends within the thalamus were thought to be linked to reduced architectural complexity of nerve fibers. γ-metrics together with conventional diffusion-metrics can more comprehensively characterize the complex mechanisms underlining age-related changes than conventional diffusion techniques alone.

Intra- and extra-axonal axial diffusivities in the white matter: Which one is faster?

A two-compartment model of diffusion in white matter, which accounts for intra- and extra-axonal spaces, is associated with two plausible mathematical scenarios: either the intra-axonal axial diffusivity D_a,‖ is higher than the extra-axonal D_e,‖ (Branch 1), or the opposite, i.e. D_a,‖ < D_e,‖ (Branch 2). This duality calls for an independent validation of compartment axial diffusivities, to determine which of the two cases holds. The aim of the present study was to use an intracerebroventricular injection of a gadolinium-based contrast agent to selectively reduce the extracellular water signal in the rat brain, and compare diffusion metrics in the genu of the corpus callosum before and after gadolinium infusion. The diffusion metrics considered were diffusion and kurtosis tensor metrics, as well as compartment-specific estimates of the WMTI-Watson two-compartment model. A strong decrease in genu T₁ and T₂ relaxation times post-Gd was observed (p < 0.001), as well as an increase of 48% in radial kurtosis (p < 0.05), which implies that the relative fraction of extracellular water signal was selectively decreased. This was further supported by a significant increase in intra-axonal water fraction as estimated from the two-compartment model, for both branches (p < 0.01 for Branch 1, p < 0.05 for Branch 2). However, pre-Gd estimates of axon dispersion in Branch 1 agreed better with literature than those of Branch 2. Furthermore, comparison of post-Gd changes in diffusivity and dispersion between data and simulations further supported Branch 1 as the biologically plausible solution, i.e. D_a,‖ > D_e,‖. This result is fully consistent with other recent measurements of compartment axial diffusivities that used entirely different approaches, such as diffusion tensor encoding.

Design and characterization of tissue-mimicking gel phantoms for diffusion kurtosis imaging

PURPOSE

The aim of this work was to create tissue-mimicking gel phantoms appropriate for diffusion kurtosis imaging (DKI) for quality assurance, protocol optimization, and sequence development.

METHODS

A range of agar, agarose, and polyvinyl alcohol phantoms with concentrations ranging from 1.0% to 3.5%, 0.5% to 3.0%, and 10% to 20%, respectively, and up to 3 g of glass microspheres per 100 ml were created. Diffusion coefficients, excess kurtosis values, and relaxation rates were experimentally determined.

RESULTS

The kurtosis values for the plain gels ranged from 0.05 with 95% confidence interval (CI) of (0.029,0.071) to 0.216(0.185,0.246), well below the kurtosis values reported in the literature for various tissues. The addition of glass microspheres increased the kurtosis of the gels with values up to 0.523(0.465,0.581) observed for gels with the highest concentration of microspheres. Repeat scans of some of the gels after more than 6 months of storage at room temperature indicate changes in the diffusion parameters of less than 10%. The addition of the glass microspheres reduces the apparent diffusion coefficients (ADCs) and increases the longitudinal and transverse relaxation rates, but the values remain comparable to those for plain gels and tissue, with ADCs observed ranging from 818(585,1053) × 10^-6  mm² /s to 2257(2118,2296) × 10^-6  mm² /s, R₁ values ranging from 0.34(0.32,0.35) 1/s to 0.51(0.50,0.52) 1/s, and R₂ values ranging from 9.69(9.34,10.04) 1/s to 33.07(27.10, 39.04) 1/s.

CONCLUSIONS

Glass microspheres can be used to effectively modify diffusion properties of gel phantoms and achieve a range of kurtosis values comparable to those reported for a variety of tissues.

Automated detection of pathologic white matter alterations in Alzheimer's disease using combined diffusivity and kurtosis method

Diffusion tensor imaging (DTI) and diffusion kurtosis imaging (DKI) are important diffusion MRI techniques for detecting microstructure abnormities in diseases such as Alzheimer's. The advantages of DKI over DTI have been reported generally; however, the indistinct relationship between diffusivity and kurtosis has not been clearly revealed in clinical settings. In this study, we hypothesize that the combination of diffusivity and kurtosis in DKI improves the capacity of DKI to detect Alzheimer's disease compared with diffusivity or kurtosis alone. Specifically, a support vector machine-based approach was applied to combine diffusivity and kurtosis and to compare different indices datasets. Strict assessments were conducted to ensure the reliability of all classifiers. Then, data from the optimized classifiers were used to detect abnormalities. With the combination, high accuracy performances of 96.23% were obtained in 53 subjects, including 27 Alzheimer's patients. More highly scored abnormal regions were selected by the combination than alone. The results revealed that more precise diffusivity and complementary kurtosis mainly contributed to the high performance of the combination in DKI. This study provides further understanding of DKI and the relationship between diffusivity and kurtosis in pathologic white matter alterations in Alzheimer's disease.

Concerning the matching of magnetic susceptibility differences for the compensation of background gradients in anisotropic diffusion fibre phantoms

Artificial, anisotropic fibre phantoms are nowadays increasingly used in the field of diffusion-weighted MRI. Such phantoms represent useful tools for, among others, the calibration of pulse sequences and validation of diffusion models since they can mimic well-known structural features of brain tissue on the one hand, but exhibit a reduced complexity, on the other. Among all materials, polyethylene fibres have been widely used due to their excellent properties regarding the restriction of water diffusion and surface relaxation properties. Yet the magnetic susceptibility of polyethylene can be distinctly lower than that of distilled water. This difference produces strong microscopic, background field gradients in the vicinity of fibre bundles which are not parallel to the static magnetic field. This, in turn, modulates the MRI signal behaviour. In the present work we investigate an approach to reduce the susceptibility-induced background gradients via reducing the heterogeneity in the internal magnetic susceptibility. An aqueous solution of magnesium chloride hexahydrate (MgCl₂·6H₂O) is used for this purpose. Its performance is demonstrated in dedicated anisotropic fibre phantoms with different geometrical configurations.

Comparison of Diffusion Metrics Obtained at 1.5T and 3T in Human Brain With Diffusional Kurtosis Imaging

PURPOSE

To quantitatively compare diffusion metrics for human brain estimated with diffusional kurtosis imaging (DKI) at applied field strengths of 1.5 and 3T.

MATERIALS AND METHODS

DKI data for brain were acquired at both 1.5 and 3T from each of six healthy volunteers using a twice-refocused diffusion-weighted imaging sequence. From these data, parametric maps of mean diffusivity (MD), axial diffusivity (D_‖ ), radial diffusivity (D_⊥ ), fractional anisotropy (FA), mean diffusional kurtosis (MK), axial kurtosis (K_‖ ), radial kurtosis (K_⊥ ), and kurtosis fractional anisotropy (KFA) were estimated. Comparisons of the results from the two field strengths were made for each metric using both Bland-Altman plots and linear regression to calculate coefficients of determination (R² ) and best fit lines.

RESULTS

Diffusion metrics measured at 1.5 and 3T were observed to be similar. Linear regression of the full datasets had coefficients of determination varying from a low of R² = 0.86 for KFA to a high of R² = 0.97 for FA. The slopes of the 3T vs. 1.5T best linear fits varied from 0.881 ± 0.009 for KFA to 1.038 ± 0.010 for D_‖ . From a Bland-Altman analysis of selected regions of interest, the mean differences of the metrics for the two field strengths were all found to be less than 6%, except for KFA, which showed the largest relative discrepancy of 10%.

CONCLUSION

Diffusion metrics measured with DKI at 1.5 and 3T are strongly correlated and typically differ by only a few percent. The somewhat higher discrepancy for the KFA is argued to mainly reflect the effects of signal noise. This supports the robustness DKI results with respect to field strength.

LEVEL OF EVIDENCE

3 J. Magn. Reson. Imaging 2017;45:673-680.