Reproducibility and repeatability of magnetic resonance imaging in dementia
Introduction
People with mild cognitive impairment (MCI) [1] show age-related decline that is greater than that of their age-matched peers. A diagnosis of dementia is given when a significant loss of everyday cognitive function, unrelated to frailty, is identified. Approximately 50 million people are living with dementia worldwide, a number set to increase three-fold by 2050 [2]. Although MCI can be a precursor, not all people with MCI go on to develop dementia [3]. The reasons why some individuals progress to dementia and others do not is unresolved and remains a topic of focussed research. Biomarkers are needed that accurately track cognitive decline and hold potential to predict which individuals will progress to dementia.
Quantitative imaging biomarkers (QIBs) assist with diagnosis, disease-monitoring, and assessment of treatment and interventions. QIBs derived from magnetic resonance imaging (MRI) that are of interest in dementia studies [4] include grey matter brain volumes derived from T1-weighted (T1w) imaging for tissue atrophy, arterial spin labelling (ASL) metrics for cerebral blood flow (hypo-perfusion/hypo-metabolism), diffusion weighted imaging (DWI) measures for white matter structure, resting state functional MRI (rsfMRI) for functional connectivity and T2w fluid attenuation inversion recovery (FLAIR) derived estimates of white matter hyperintensity (WMH) volume for small vessel disease. Additional, more clinically focused scans in a dementia protocol often include T2-weighted (T2w) and susceptibility weighted imaging (SWI) to assess vascular health and other pathologies.
Multicentre studies allow for larger, more representative cohorts to be recruited for research trials. However, inter-site measurement variability needs to be quantifiable to interpret pooled data. For QIBs to be adopted widely and incorporated into clinical practice, inter-site measurement reliability needs to be established to determine confidence in diagnostic metrics. Knowledge of QIB variability over time due to measurement uncertainty is essential when monitoring longitudinal cognitive changes as a function of normal ageing, disease, or an intervention. Without this crucial information, it cannot be determined whether subsequent changes in the QIB are due to underlying physiological changes, or simply measurement variability.
The Quantitative Imaging Biomarkers Alliance (QIBA) [5], [6] defines reproducibility as variability due to measurements being collected under different conditions, e.g., at different sites with different hardware or processed with different software. Conversely repeatability refers to variability in a measurement collected under the same conditions multiple times. Several studies have examined the reproducibility and repeatability of individual, or single modality QIBs within the context of dementia [7], [8], [9]. However multiple QIBs in combination, akin to a “biomarker signature”, are likely to have more predictive power of cognitive decline than a single maker alone [10]. Accordingly, we wished to assess measurement variability in a suite of parameters that could be used for this purpose. We recruited “travelling heads” (THs), the same participants who travelled to imaging centres to be scanned at repeated time-points, enabling assessment of the reproducibility and repeatability of quantitative MRI (qMRI) markers for dementia. Each site was part of a multicentre, longitudinal study known as the Dementia Prevention Research Clinics.
Section snippets
Participants
Six participants (3 female, 3 male) were recruited as THs. Their average age at commencement of the study was 38.5 years (range 31.9–52.7). The study received ethical approval from the Health and Disability Ethics Committee and all participants provided informed written consent before taking part, per New Zealand National Ethical Standards.
Data acquisition
Imaging was performed at three different cities in New Zealand: Auckland, Christchurch, and Dunedin. Each site has a similar 3T MRI system (MAGNETOM Skyra,
Results
From a total of 266 scans (38 scan sessions × 7 modalities), 4 scans were missing/unusable. One baseline ASL scan was excluded due to poor labelling, evident by very low perfusion-weighted signal. For this participant, repeatability ASL measures were normalised to their second ASL scan. For another participant, the FLAIR, T2w, and SWI scans were not collected due to limited scan time availability. Representative images from a single participant scanned at the three different sites within seven
Discussion
Overall, we found that 15 QIBs derived from MRI modalities typically found in a dementia imaging protocol [4], were comparable across all sites participating in our Dementia Prevention Research Clinics. Generally, reproducibility of metrics derived from brain structure (e.g., tissue volume) was better than those measuring physiological brain processes (e.g., resting state connectivity and perfusion). For a given metric, reproducibility CoV was of similar magnitude to long-term repeatability
Conclusion
In this work, we investigated the reproducibility (inter-site) and repeatability (intra-site, over both short (days) and long (years) time periods) of 15 quantitative MRI metrics in the context of an ongoing longitudinal investigation of MCI and dementia. Structural metrics exhibited excellent reproducibility across three sites and repeatability over both days and up to five years. Resting state fMRI showed poorer reproducibility and repeatability, while perfusion MRI showed intermediate
Funding source
This study was funded in whole by Brain Research New Zealand - Rangahau Roro Aotearoa, a government funded Centre of Research Excellence. DRA receives salary support from the Canada 150 Research Program.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
The authors thank the travelling head participants for their time and dedication to the study. We thank the imaging staff of the Centre for Advanced Magnetic Resonance Imaging (Auckland) and Pacific Radiology Group (Dunedin and Christchurch) for MRI scanning, and the Centre for eResearch, The University of Auckland, for computing support. The authors acknowledge Siemens Healthcare for the provision of a 3D pCASL prototype sequence and Dr. Marta Vidorreta De Cerio and Josef Pfeuffer for advice
References (75)
- et al.
Dementia prevention, intervention, and care: 2020 report of the Lancet Commission
Lancet
(2020) - et al.
Quantitative MRI: Defining repeatability, reproducibility and accuracy for prostate cancer imaging biomarker development
Magn Reson Imaging
(2021) - et al.
Image processing and Quality Control for the first 10,000 brain imaging datasets from UK Biobank
Neuroimage
(2018) - et al.
Unified segmentation
Neuroimage
(2005) - et al.
A Probabilistic Atlas of the Human Brain: Theory and Rationale for Its Development: The International Consortium for Brain Mapping (ICBM)
NeuroImage
(1995) - et al.
Fast and robust parameter estimation for statistical partial volume models in brain MRI
Neuroimage
(2004) - et al.
Cerebral atrophy in mild cognitive impairment: A systematic review with meta-analysis
Alzheimer’s Dementia: Diagnosis Assessment Disease Monitor
(2015) - et al.
Automated anatomical labelling atlas 3
NeuroImage
(2020) - et al.
Automated Anatomical Labeling of Activations in SPM Using a Macroscopic Anatomical Parcellation of the MNI MRI Single-Subject Brain
NeuroImage
(2002) - et al.
ICA-AROMA: A robust ICA-based strategy for removing motion artifacts from fMRI data
Neuroimage
(2015)
Complex network measures of brain connectivity: Uses and interpretations
NeuroImage
Comparison of 2D and 3D single-shot ASL perfusion fMRI sequences
NeuroImage
Review FSL
NeuroImage
A reproducible evaluation of ANTs similarity metric performance in brain image registration
NeuroImage
Complex diffusion-weighted image estimation via matrix recovery under general noise models
NeuroImage
Denoising of diffusion MRI using random matrix theory
NeuroImage
An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging
NeuroImage
Advances in functional and structural MR image analysis and implementation as FSL
NeuroImage
Interpolation of diffusion weighted imaging datasets
NeuroImage
Multi-tissue constrained spherical deconvolution for improved analysis of multi-shell diffusion MRI data
NeuroImage
Investigating white matter fibre density and morphology using fixel-based analysis
NeuroImage
Symmetric diffeomorphic registration of fibre orientation distributions
NeuroImage
Apparent Fibre Density: A novel measure for the analysis of diffusion-weighted magnetic resonance images
NeuroImage
Spherical-deconvolution informed filtering of tractograms
NeuroImage
Connectivity-based fixel enhancement: Whole-brain statistical analysis of diffusion MRI measures in the presence of crossing fibres
Neuroimage
An automated tool for detection of FLAIR-hyperintense white-matter lesions in Multiple Sclerosis
NeuroImage
On visualizing phonetic data from repeated measures experiments with multiple random effects
J Phonetics
Harmonizing brain magnetic resonance imaging methods for vascular contributions to neurodegeneration
Alzheimers Dement (Amst)
The topography of white matter hyperintensities on brain MRI in healthy 60- to 64-year-old individuals
NeuroImage
A meta-analysis of hippocampal atrophy rates in Alzheimer’s disease
Neurobiol Aging
A decade of test-retest reliability of functional connectivity: A systematic review and meta-analysis
Neuroimage
Striving toward translation: strategies for reliable fMRI measurement
Trends Cogn Sci
Mild cognitive impairment as a diagnostic entity
J Intern Med
Practice guideline update summary: Mild cognitive impairment: Report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology
Neurology
Dementia imaging in clinical practice: a European-wide survey of 193 centres and conclusions by the ESNR working group
Neuroradiology
Quantitative imaging biomarkers: A review of statistical methods for technical performance assessment
Stat Methods Med Res
Multiparameter mapping of relaxation (R1, R2*), proton density and magnetization transfer saturation at 3 T: A multicenter dual-vendor reproducibility and repeatability study
Hum Brain Mapp
Cited by (3)
Image restoration algorithm incorporating methods to remove noise and blurring from positron emission tomography imaging for Alzheimer's disease diagnosis
2022, Physica MedicaCitation Excerpt :Although the etiology of this disease has not yet been clearly elucidated, it is known that extensive neuronal destruction occurs due to deposition of beta-amyloid protein as a consequence of its increased production and decreased excretion [18,19]. Although it is important to subjectively judge clinical symptoms for the diagnosis of AD, quantitative evaluation using the image such as MRI and PET can secure reproducibility and repeatability in determining AD diagnosis [20]. Moreover, it is essential to acquire a PET image to quantitatively analyze the structural and functional changes that occur in the brain.
Evaluation of gauges in measurement systems
2023, Production Engineering