Validation of DWI pre-processing procedures for reliable differentiation between human brain gliomas

What's new and what's next in diffusion MRI preprocessing

Diffusion MRI (dMRI) provides invaluable information for the study of tissue microstructure and brain connectivity, but suffers from a range of imaging artifacts that greatly challenge the analysis of results and their interpretability if not appropriately accounted for. This review will cover dMRI artifacts and preprocessing steps, some of which have not typically been considered in existing pipelines or reviews, or have only gained attention in recent years: brain/skull extraction, B-matrix incompatibilities w.r.t the imaging data, signal drift, Gibbs ringing, noise distribution bias, denoising, between- and within-volumes motion, eddy currents, outliers, susceptibility distortions, EPI Nyquist ghosts, gradient deviations, B1 bias fields, and spatial normalization. The focus will be on "what's new" since the notable advances prior to and brought by the Human Connectome Project (HCP), as presented in the predecessing issue on "Mapping the Connectome" in 2013. In addition to the development of novel strategies for dMRI preprocessing, exciting progress has been made in the availability of open source tools and reproducible pipelines, databases and simulation tools for the evaluation of preprocessing steps, and automated quality control frameworks, amongst others. Finally, this review will consider practical considerations and our view on "what's next" in dMRI preprocessing.

Memory recovery in relation to default mode network impairment and neurite density during brain tumor treatment

OBJECTIVE

The aim of this study was to test brain tumor interactions with brain networks, thereby identifying protective features and risk factors for memory recovery after resection.

METHODS

Seventeen patients with diffuse nonenhancing glioma (ages 22-56 years) underwent longitudinal MRI before and after surgery, and during a 12-month recovery period (47 MRI scans in total after exclusion). After each scanning session, a battery of memory tests was performed using a tablet-based screening tool, including free verbal memory, overall verbal memory, episodic memory, orientation, forward digit span, and backward digit span. Using structural MRI and neurite orientation dispersion and density imaging (NODDI) derived from diffusion-weighted images, the authors estimated lesion overlap and neurite density, respectively, with brain networks derived from normative data in healthy participants (somatomotor, dorsal attention, ventral attention, frontoparietal, and default mode network [DMN]). Linear mixed-effect models (LMMs) that regressed out the effect of age, gender, tumor grade, type of treatment, total lesion volume, and total neurite density were used to test the potential longitudinal associations between imaging markers and memory recovery.

RESULTS

Memory recovery was not significantly associated with either the tumor location based on traditional lobe classification or the type of treatment received by patients (i.e., surgery alone or surgery with adjuvant chemoradiotherapy). Nonlocal effects of tumors were evident on neurite density, which was reduced not only within the tumor but also beyond the tumor boundary. In contrast, high preoperative neurite density outside the tumor but within the DMN was associated with better memory recovery (LMM, p value after false discovery rate correction [Pfdr] < 10-3). Furthermore, postoperative and follow-up neurite density within the DMN and frontoparietal network were also associated with memory recovery (LMM, Pfdr = 0.014 and Pfdr = 0.001, respectively). Preoperative tumor and postoperative lesion overlap with the DMN showed a significant negative association with memory recovery (LMM, Pfdr = 0.002 and Pfdr < 10-4, respectively).

CONCLUSIONS

Imaging biomarkers of cognitive recovery and decline can be identified using NODDI and resting-state networks. Brain tumors and their corresponding treatment affecting brain networks that are fundamental for memory functioning such as the DMN can have a major impact on patients' memory recovery.

Feasibility of generalised diffusion kurtosis imaging approach for brain glioma grading

Purpose

An accurate differentiation of brain glioma grade constitutes an important clinical issue. Powerful non-invasive approach based on diffusion MRI has already demonstrated its feasibility in glioma grade stratification. However, the conventional diffusion tensor (DTI) and kurtosis imaging (DKI) demonstrated moderate sensitivity and performance in glioma grading. In the present work, we apply generalised DKI (gDKI) approach in order to assess its diagnostic accuracy and potential application in glioma grading.

Methods

Diffusion scalar metrics were obtained from 50 patients with different glioma grades confirmed by histological tests following biopsy or surgery. All patients were divided into two groups with low- and high-grade gliomas as grade II versus grades III and IV, respectively. For a comparison, trained radiologists segmented the brain tissue into three regions with solid tumour, oedema, and normal appearing white matter. For each region, we estimated the conventional and gDKI metrics including DTI maps.

Results

We found high correlations between DKI and gDKI metrics in high-grade glioma. Further, gDKI metrics enabled introduction of a complementary measure for glioma differentiation based on correlations between the conventional and generalised approaches. Both conventional and generalised DKI metrics showed quantitative maps of tumour heterogeneity and oedema behaviour. gDKI approach demonstrated largely similar sensitivity and specificity in low-high glioma differentiation as in the case of conventional DKI method.

Conclusion

The generalised diffusion kurtosis imaging enables differentiation of low- and high-grade gliomas at the same level as the conventional DKI. Additionally, gDKI exhibited higher sensitivity to tumour heterogeneity and tissue contrast between tumour and healthy tissue and, thus, may contribute as a complementary source of information on tumour differentiation.

Towards an optimised processing pipeline for diffusion magnetic resonance imaging data: Effects of artefact corrections on diffusion metrics and their age associations in UK Biobank

Increasing interest in the structural and functional organisation of the human brain encourages the acquisition of big data sets comprising multiple neuroimaging modalities, often accompanied by additional information obtained from health records, cognitive tests, biomarkers and genotypes. Diffusion weighted magnetic resonance imaging data enables a range of promising imaging phenotypes probing structural connections as well as macroanatomical and microstructural properties of the brain. The reliability and biological sensitivity and specificity of diffusion data depend on processing pipeline. A state-of-the-art framework for data processing facilitates cross-study harmonisation and reduces pipeline-related variability. Using diffusion magnetic resonance imaging (MRI) data from 218 individuals in the UK Biobank, we evaluate the effects of different processing steps that have been suggested to reduce imaging artefacts and improve reliability of diffusion metrics. In lack of a ground truth, we compared diffusion metric sensitivity to age between pipelines. By comparing distributions and age sensitivity of the resulting diffusion metrics based on different approaches (diffusion tensor imaging, diffusion kurtosis imaging and white matter tract integrity), we evaluate a general pipeline comprising seven postprocessing blocks: noise correction; Gibbs ringing correction; evaluation of field distortions; susceptibility, eddy-current and motion-induced distortion corrections; bias field correction; spatial smoothing and final diffusion metric estimations. Based on this evaluation, we suggest an optimised processing pipeline for diffusion weighted MRI data.

Feasibility of Non-Gaussian Diffusion Metrics in Chronic Disorders of Consciousness

Diagnostic accuracy of different chronic disorders of consciousness (DOC) can be affected by the false negative errors in up to 40% cases. In the present study, we aimed to investigate the feasibility of a non-Gaussian diffusion approach in chronic DOC and to estimate a sensitivity of diffusion kurtosis imaging (DKI) metrics for the differentiation of vegetative state/unresponsive wakefulness syndrome (VS/UWS) and minimally conscious state (MCS) from a healthy brain state. We acquired diffusion MRI data from 18 patients in chronic DOC (11 VS/UWS, 7 MCS) and 14 healthy controls. A quantitative comparison of the diffusion metrics for grey (GM) and white (WM) matter between the controls and patient group showed a significant (p < 0.05) difference in supratentorial WM and GM for all evaluated diffusion metrics, as well as for brainstem, corpus callosum, and thalamus. An intra-subject VS/UWS and MCS group comparison showed only kurtosis metrics and fractional anisotropy differences using tract-based spatial statistics, owing mainly to macrostructural differences on most severely lesioned hemispheres. As a result, we demonstrated an ability of DKI metrics to localise and detect changes in both WM and GM and showed their capability in order to distinguish patients with a different level of consciousness.

Threshold Isocontouring on High b-Value Diffusion-Weighted Images in Magnetic Resonance Mammography

OBJECTIVES

Motivated by the similar appearance of malignant breast lesions in high b-value diffusion-weighted imaging (DWI) and positron emission tomography, the purpose of this work was to evaluate the applicability of a threshold isocontouring approach commonly used in positron emission tomography to analyze DWI data acquired from female human breasts with minimal interobserver variability.

METHODS

Twenty-three female participants (59.4 ± 10.0 years) with 23 lesions initially classified as suggestive of cancers in x-ray mammography screening were subsequently imaged on a 1.5-T magnetic resonance imaging scanner. Diffusion-weighted imaging was performed prior to biopsy with b values of 0, 100, 750, and 1500 s/mm. Isocontouring with different threshold levels was performed on the highest b-value image to determine the voxels used for subsequent evaluation of diffusion metrics. The coefficient of variation was computed by specifying 4 different regions of interest drawn around the lesion. Additionally, a receiver operating statistical analysis was performed.

RESULTS

Using a relative threshold level greater than or equal to 0.85 almost completely suppresses the intra-individual and inter-individual variability. Among 4 studied diffusion metrics, the diffusion coefficients from the intravoxel incoherent motion model returned the highest area under curve value of 0.9. The optimal cut-off diffusivity was found to be 0.85 μm/ms with a sensitivity of 87.5% and specificity of 90.9%.

CONCLUSION

Threshold isocontouring on high b-value maps is a viable approach to reliably evaluate DWI data of suspicious focal lesions in magnetic resonance mammography.

Differentiation of glioma malignancy grade using diffusion MRI

Modern diffusion MR protocols allow one to acquire the multi-shell diffusion data with high diffusion weightings in a clinically feasible time. In the present work we assessed three diffusion approaches based on diffusion and kurtosis tensor imaging (DTI, DKI), and neurite orientation dispersion and density imaging (NODDI) as possible biomarkers for human brain glioma grade differentiation based on the one diffusion protocol. We used three diffusion weightings (so called b-values) equal to 0, 1000, and 2500s/mm² with 60 non-coplanar diffusion directions in the case of non-zero b-values. The patient groups of the glioma grades II, III, and IV consist of 8 subjects per group. We found that DKI, and NODDI scalar metrics can be effectively used as glioma grade biomarkers with a significant difference (p<0.05) for grading between low- and high-grade gliomas, in particular, for glioma II versus glioma III grades, and glioma III versus glioma IV grades. The use of mean/axial kurtosis and intra-axonal fraction/orientation dispersion index metrics allowed us to obtain the most feasible and reliable differentiation criteria. For example, in the case of glioma grades II, III, and IV the mean kurtosis is equal to 0.31, 0.51, and 0.90, and the orientation dispersion index is equal to 0.14, 0.30, and 0.59, respectively. The limitations and perspectives of the biophysical diffusion models based on intra-/extra-axonal compartmentalisation for glioma differentiation are discussed.