Search

Deep learning (DL) has shown great potential in medical image enhancement problems, such as super-resolution or image synthesis. However, to date, most existing approaches are based on deterministic models, neglecting the presence of different sources of uncertainty in such problems. Here we introduce methods to characterise different components of uncertainty, and demonstrate the ideas using diffusion MRI super-resolution. Specifically, we propose to account for intrinsic uncertainty through a heteroscedastic noise model and for parameter uncertainty through approximate Bayesian inference, and integrate the two to quantify predictive uncertainty over the output image. Moreover, we introduce a method to propagate the predictive uncertainty on a multi-channelled image to derived scalar parameters, and separately quantify the effects of intrinsic and parameter uncertainty therein. The methods are evaluated for super-resolution of two different signal representations of diffusion MR images-Diffusion Tensor images and Mean Apparent Propagator MRI-and their derived quantities such as mean diffusivity and fractional anisotropy, on multiple datasets of both healthy and pathological human brains. Results highlight three key potential benefits of modelling uncertainty for improving the safety of DL-based image enhancement systems. Firstly, modelling uncertainty improves the predictive performance even when test data departs from training data ("out-of-distribution" datasets). Secondly, the predictive uncertainty highly correlates with reconstruction errors, and is therefore capable of detecting predictive "failures". Results on both healthy subjects and patients with brain glioma or multiple sclerosis demonstrate that such an uncertainty measure enables subject-specific and voxel-wise risk assessment of the super-resolved images that can be accounted for in subsequent analysis. Thirdly, we show that the method for decomposing predictive uncertainty into its independent sources provides high-level "explanations" for the model performance by separately quantifying how much uncertainty arises from the inherent difficulty of the task or the limited training examples. The introduced concepts of uncertainty modelling extend naturally to many other imaging modalities and data enhancement applications., (Crown Copyright © 2020. Published by Elsevier Inc. All rights reserved.)

Cross-scanner and cross-protocol variability of diffusion magnetic resonance imaging (dMRI) data are known to be major obstacles in multi-site clinical studies since they limit the ability to aggregate dMRI data and derived measures. Computational algorithms that harmonize the data and minimize such variability are critical to reliably combine datasets acquired from different scanners and/or protocols, thus improving the statistical power and sensitivity of multi-site studies. Different computational approaches have been proposed to harmonize diffusion MRI data or remove scanner-specific differences. To date, these methods have mostly been developed for or evaluated on single b-value diffusion MRI data. In this work, we present the evaluation results of 19 algorithms that are developed to harmonize the cross-scanner and cross-protocol variability of multi-shell diffusion MRI using a benchmark database. The proposed algorithms rely on various signal representation approaches and computational tools, such as rotational invariant spherical harmonics, deep neural networks and hybrid biophysical and statistical approaches. The benchmark database consists of data acquired from the same subjects on two scanners with different maximum gradient strength (80 and 300 ​mT/m) and with two protocols. We evaluated the performance of these algorithms for mapping multi-shell diffusion MRI data across scanners and across protocols using several state-of-the-art imaging measures. The results show that data harmonization algorithms can reduce the cross-scanner and cross-protocol variabilities to a similar level as scan-rescan variability using the same scanner and protocol. In particular, the LinearRISH algorithm based on adaptive linear mapping of rotational invariant spherical harmonics features yields the lowest variability for our data in predicting the fractional anisotropy (FA), mean diffusivity (MD), mean kurtosis (MK) and the rotationally invariant spherical harmonic (RISH) features. But other algorithms, such as DIAMOND, SHResNet, DIQT, CMResNet show further improvement in harmonizing the return-to-origin probability (RTOP). The performance of different approaches provides useful guidelines on data harmonization in future multi-site studies., (Copyright © 2020. Published by Elsevier Inc.)

Multi-parametric quantitative MRI (qMRI) of the spinal cord is a promising non-invasive tool to probe early microstructural damage in neurological disorders. It is usually performed in vivo by combining acquisitions with multiple signal readouts, which exhibit different thermal noise levels, geometrical distortions and susceptibility to physiological noise. This ultimately hinders joint multi-contrast modelling and makes the geometric correspondence of parametric maps challenging. We propose an approach to overcome these limitations, by implementing state-of-the-art microstructural MRI of the spinal cord with a unified signal readout in vivo (i.e. with matched spatial encoding parameters across a range of imaging contrasts). We base our acquisition on single-shot echo planar imaging with reduced field-of-view, and obtain data from two different vendors (vendor 1: Philips Achieva; vendor 2: Siemens Prisma). Importantly, the unified acquisition allows us to compare signal and noise across contrasts, thus enabling overall quality enhancement via multi-contrast image denoising methods. As a proof-of-concept, here we provide a demonstration with one such method, known as Marchenko-Pastur (MP) Principal Component Analysis (PCA) denoising. MP-PCA is a singular value (SV) decomposition truncation approach that relies on redundant acquisitions, i.e. such that the number of measurements is large compared to the number of components that are maintained in the truncated SV decomposition. Here we used in vivo and synthetic data to test whether a unified readout enables more efficient MP-PCA denoising of less redundant acquisitions, since these can be denoised jointly with more redundant ones. We demonstrate that a unified readout provides robust multi-parametric maps, including diffusion and kurtosis tensors from diffusion MRI, myelin metrics from two-pool magnetisation transfer, and T1 and T2 from relaxometry. Moreover, we show that MP-PCA improves the quality of our multi-contrast acquisitions, since it reduces the coefficient of variation (i.e. variability) by up to 17% for mean kurtosis, 8% for bound pool fraction (myelin-sensitive), and 13% for T1, while enabling more efficient denoising of modalities limited in redundancy (e.g. relaxometry). In conclusion, multi-parametric spinal cord qMRI with unified readout is feasible and provides robust microstructural metrics with matched resolution and distortions, whose quality benefits from multi-contrast denoising methods such as MP-PCA., Competing Interests: Declaration of competing interests TS is an employee of Philips UK. EF, JV, DSN and NYU School of Medicine, are co-inventors in the MP-PCA technology related to this research; a patent application has been filed and is pending. EF, JV, DSN, and TMS are shareholders and hold advisory roles at Microstructure Imaging, Inc., (Copyright © 2020 The Authors. Published by Elsevier Inc. All rights reserved.)

Spinal cord atrophy measurements obtained from structural magnetic resonance imaging (MRI) are associated with disability in many neurological diseases and serve as in vivo biomarkers of neurodegeneration. Longitudinal spinal cord atrophy rate is commonly determined from the numerical difference between two volumes (based on 3D surface fitting) or two cross-sectional areas (CSA, based on 2D edge detection) obtained at different time-points. Being an indirect measure, atrophy rates are susceptible to variable segmentation errors at the edge of the spinal cord. To overcome those limitations, we developed a new registration-based pipeline that measures atrophy rates directly. We based our approach on the generalised boundary shift integral (GBSI) method, which registers 2 scans and uses a probabilistic XOR mask over the edge of the spinal cord, thereby measuring atrophy more accurately than segmentation-based techniques. Using a large cohort of longitudinal spinal cord images (610 subjects with multiple sclerosis from a multi-centre trial and 52 healthy controls), we demonstrated that GBSI is a sensitive, quantitative and objective measure of longitudinal spinal cord volume change. The GBSI pipeline is repeatable, reproducible, and provides more precise measurements of longitudinal spinal cord atrophy than segmentation-based methods in longitudinal spinal cord atrophy studies., (Copyright © 2020 The Authors. Published by Elsevier Inc. All rights reserved.)

Diffusion MRI is being used increasingly in studies of the brain and other parts of the body for its ability to provide quantitative measures that are sensitive to changes in tissue microstructure. However, inter-scanner and inter-protocol differences are known to induce significant measurement variability, which in turn jeopardises the ability to obtain 'truly quantitative measures' and challenges the reliable combination of different datasets. Combining datasets from different scanners and/or acquired at different time points could dramatically increase the statistical power of clinical studies, and facilitate multi-centre research. Even though careful harmonisation of acquisition parameters can reduce variability, inter-protocol differences become almost inevitable with improvements in hardware and sequence design over time, even within a site. In this work, we present a benchmark diffusion MRI database of the same subjects acquired on three distinct scanners with different maximum gradient strength (40, 80, and 300 mT/m), and with 'standard' and 'state-of-the-art' protocols, where the latter have higher spatial and angular resolution. The dataset serves as a useful testbed for method development in cross-scanner/cross-protocol diffusion MRI harmonisation and quality enhancement. Using the database, we compare the performance of five different methods for estimating mappings between the scanners and protocols. The results show that cross-scanner harmonisation of single-shell diffusion data sets can reduce the variability between scanners, and highlight the promises and shortcomings of today's data harmonisation techniques., (Copyright © 2019. Published by Elsevier Inc.)

An important image processing step in spinal cord magnetic resonance imaging is the ability to reliably and accurately segment grey and white matter for tissue specific analysis. There are several semi- or fully-automated segmentation methods for cervical cord cross-sectional area measurement with an excellent performance close or equal to the manual segmentation. However, grey matter segmentation is still challenging due to small cross-sectional size and shape, and active research is being conducted by several groups around the world in this field. Therefore a grey matter spinal cord segmentation challenge was organised to test different capabilities of various methods using the same multi-centre and multi-vendor dataset acquired with distinct 3D gradient-echo sequences. This challenge aimed to characterize the state-of-the-art in the field as well as identifying new opportunities for future improvements. Six different spinal cord grey matter segmentation methods developed independently by various research groups across the world and their performance were compared to manual segmentation outcomes, the present gold-standard. All algorithms provided good overall results for detecting the grey matter butterfly, albeit with variable performance in certain quality-of-segmentation metrics. The data have been made publicly available and the challenge web site remains open to new submissions. No modifications were introduced to any of the presented methods as a result of this challenge for the purposes of this publication., (Copyright © 2017 The Authors. Published by Elsevier Inc. All rights reserved.)

Here we present the application of neurite orientation dispersion and density imaging (NODDI) to the healthy spinal cord in vivo. NODDI provides maps such as the intra-neurite tissue volume fraction (vin), the orientation dispersion index (ODI) and the isotropic volume fraction (viso), and here we investigate their potential for spinal cord imaging. We scanned five healthy volunteers, four of whom twice, on a 3T MRI system with a ZOOM-EPI sequence. In accordance to the published NODDI protocol, multiple b-shells were acquired at cervical level and both NODDI and diffusion tensor imaging (DTI) metrics were obtained and analysed to: i) characterise differences in grey and white matter (GM/WM); ii) assess the scan-rescan reproducibility of NODDI; iii) investigate the relationship between NODDI and DTI; and iv) compare the quality of fit of NODDI and DTI. Our results demonstrated that: i) anatomical features can be identified in NODDI maps, such as clear contrast between GM and WM in ODI; ii) the variabilities of vin and ODI are comparable to that of DTI and are driven by biological differences between subjects for ODI, have similar contribution from measurement errors and biological variation for vin, whereas viso shows higher variability, driven by measurement errors; iii) NODDI identifies potential sources contributing to DTI indices, as in the brain; and iv) NODDI outperforms DTI in terms of quality of fit. In conclusion, this work shows that NODDI is a useful model for in vivo diffusion MRI of the spinal cord, providing metrics closely related to tissue microstructure, in line with findings in the brain., (Copyright © 2015. Published by Elsevier Inc.)