Improved reliability of hippocampal atrophy rate measurement in mild cognitive impairment using fluid registration

doi:10.1016/j.neuroimage.2006.10.033

NeuroImage

Volume 34, Issue 3, 1 February 2007, Pages 1036-1041

https://doi.org/10.1016/j.neuroimage.2006.10.033 Get rights and content

Abstract

MRI-derived rates of hippocampal atrophy may serve as surrogate markers of disease progression in mild cognitive impairment (MCI). Manual delineation is the gold standard in hippocampal volumetry; however, this technique is time-consuming and subject to errors. We aimed to compare regional non-linear (fluid) registration measurement of hippocampal atrophy rates against manual delineation in MCI. Hippocampi of 18 subjects were manually outlined twice on MRI scan-pairs (interval ± SD: 2.01 ± 0.11 years), and volumes were subtracted to calculate change over time. Following global affine and local rigid registration, regional fluid registration was performed from which atrophy rates were derived from the Jacobian determinants over the hippocampal region. Atrophy rates as derived by fluid registration were computed using both forward (repeat onto baseline) and backward (baseline onto repeat) registration. Reliability for both methods and agreement between methods was assessed. Mean ± SD hippocampal atrophy rates (%/year) derived by manual delineation were: left: 2.13 ± 1.62; right: 2.36 ± 1.78 and for regional fluid registration: forward: left: 2.39 ± 1.68; right: 2.49 ± 1.52 and backward: left: 2.21 ± 1.51; right: 2.42 ± 1.49. Mean hippocampal atrophy rates did not differ between both methods. Reliability for manual hippocampal volume measurements (cross-sectional) was high (intraclass correlation coefficient (ICC): baseline and follow-up, left and right, > 0.99). However, the resulting ICC for manual measurements of hippocampal volume change (longitudinal) was considerably lower (left: 0.798; right: 0.850) compared with regional fluid registration (forward: left: 0.985; right: 0.988 and backward: left: 0.975; right: 0.989). We conclude that regional fluid registration is more reliable than manual delineation in assessing hippocampal atrophy rates, without sacrificing sensitivity to change. This method may be useful to quantify hippocampal volume change, given the reduction in operator time and improved precision.

Introduction

The hippocampus is one of the earliest structures to be affected by pathological changes in Alzheimer’s disease (AD) (Braak and Braak, 1991). A characteristic feature associated with AD is hippocampal atrophy. Rates of hippocampal atrophy, derived from serial MRI scans, are elevated in AD patients compared with controls, even in early stages of the disease, and correlate with clinical disease progression (Du et al., 2004, Jack et al., 1998, Jack et al., 2004).

Mild cognitive impairment (MCI) refers to non-demented individuals who show isolated impairment in one cognitive domain. Annually, about 10–15% of these individuals will decline to dementia, predominantly AD, compared with controls who convert at a rate of 1–2% (Petersen et al., 2001). In MCI, hippocampal atrophy rates fall between rates observed in AD patients and controls (Jack et al., 2000, Jack et al., 2004, Cardenas et al., 2003), and have been shown to be predictive of memory decline (Mungas et al., 2005). Subjects with MCI who subsequently progress to AD have higher hippocampal atrophy rates than those who remain cognitively stable (Jack et al., 2000, Jack et al., 2004).

Typically, manual delineation of the hippocampus using a region of interest (ROI) approach is considered the gold standard to determine hippocampal atrophy rates. However, ROI measurements are time-consuming, require skilled operators and are subject to errors in delineation. Moreover, uncorrelated errors in hippocampal outlining on serial scans may further decrease reliability in measurements of hippocampal volume change.

More time-efficient and reliable techniques to measure hippocampal volume change on MRI are needed, especially for use in clinical trials with large numbers of subjects and multiple MRI scans. Regional fluid registration of serial MRI has been proposed to be useful in tracking longitudinal changes in brain structures (Crum et al., 2001, Freeborough and Fox, 1998). This procedure uses non-linear registration, based on the physical model of a viscous fluid, to match serial MRI and models structural change over time in a predefined region (Crum et al., 2001).

In this study, we applied regional fluid registration to determine hippocampal atrophy rates in subjects with MCI, selected from a recent multi-centre, randomised, clinical trial. We aimed to compare this technique with the current gold standard, manual hippocampal delineation.

Section snippets

Subjects

MRI scans of 19 subjects with MCI were selected from a multi-centre, randomised, clinical trial (Gal-Int-11, Johnson and Johnson Pharmaceutical Research and Development, L.L.C., Titusville, NJ, USA), stratified for different levels of severity of hippocampal atrophy, as assessed by the medial temporal lobe atrophy visual rating scale (0–4) (Scheltens et al., 1992). Gal-Int-11 is a 24-month, randomised, double-blind placebo-controlled clinical trial studying the effect of galantamine on

Results

Table 1 shows the mean annualised hippocampal atrophy rates for the left and the right side as derived using each technique. The means of the two hippocampal atrophy rates as derived by forward fluid registration were 6–12% higher than the corresponding rates from the manual measurements. However, these differences were not significant (left: t = 0.94, p = 0.36; right: t = 0.52, p = 0.61). Furthermore, there was no evidence of a difference in variance in these atrophy rates (left: p = 0.81; right: p =

Discussion

This methodological study aimed to compare regional fluid registration in its ability to measure hippocampal atrophy rates, with the gold standard, manual delineation, in a sample of MCI subjects taken from a completed clinical trial. The main finding of our study is that regional fluid registration proved to be more reliable than manual delineation in assessing hippocampal atrophy rates, without sacrificing sensitivity to change.

Most previous studies investigating rates of hippocampal atrophy

Acknowledgments

L.A. van de Pol received funding from the Image Analysis Centre at the VU Medical Centre, and the European Federation of Neurological Societies. Additional funding was supplied by the Stichting Alzheimer and Neuropsychiatric Foundation.

N.C. Fox and J. Barnes acknowledge support from the UK Medical Research Council (G116/143). R.I. Scahill was supported by a UK Medical Research Council Grant (G90/86). The funding bodies had no influence on designs and interpretation of the study.

We would like to

References (26)

J. Barnes et al.
NeuroImage
(2004)
V.A. Cardenas et al.
Neurobiol. Aging
(2003)
W.R. Crum et al.
NeuroImage
(2001)
L. deToledo-Morrell et al.
Neurobiol. Aging
(2004)
P.A. Freeborough et al.
Comput. Methods Programs Biomed.
(1997)
P.M. Thompson et al.
NeuroImage
(2004)
L. Wang et al.
NeuroImage
(2003)
J.M. Bland et al.
Lancet
(1986)
H. Braak et al.
Acta Neuropathol. (Berl.)
(1991)
A.T. Du et al.
Neurology
(2004)

P.A. Freeborough et al.

J. Comput. Assist. Tomogr.

(1998)

C.R. Jack

Epilepsia

(1994)

C.R. Jack et al.

Neurology

(1998)

Cited by (37)

Non-linear registration improves statistical power to detect hippocampal atrophy in aging and dementia
2019, NeuroImage: Clinical
Citation Excerpt :
However, in all groups (CTRL, MCI and AD) and for all methods we also measured unexpected hippocampal volume increase for a few subjects (Fig. 3). Unexpected hippocampal volume increase for a few subjects can also be observed in other studies in which hippocampal atrophy rates were measured [Cover et al., 2016; Mulder et al., 2014; van de Pol et al., 2007a; Yushkevich et al., 2010]. We did not further investigate these subjects with positive PVCs, but most probably they are due to noise in the images.
To compare the performance of different methods for determining hippocampal atrophy rates using longitudinal MRI scans in aging and Alzheimer's disease (AD).
Quantifying hippocampal atrophy caused by neurodegenerative diseases is important to follow the course of the disease. In dementia, the efficacy of new therapies can be partially assessed by measuring their effect on hippocampal atrophy. In radiotherapy, the quantification of radiation-induced hippocampal volume loss is of interest to quantify radiation damage. We evaluated plausibility, reproducibility and sensitivity of eight commonly used methods to determine hippocampal atrophy rates using test-retest scans.
Manual, FSL-FIRST, FreeSurfer, multi-atlas segmentation (MALF) and non-linear registration methods (Elastix, NiftyReg, ANTs and MIRTK) were used to determine hippocampal atrophy rates on longitudinal T1-weighted MRI from the ADNI database. Appropriate parameters for the non-linear registration methods were determined using a small training dataset (N = 16) in which two-year hippocampal atrophy was measured using test-retest scans of 8 subjects with low and 8 subjects with high atrophy rates. On a larger dataset of 20 controls, 40 mild cognitive impairment (MCI) and 20  AD patients, one-year hippocampal atrophy rates were measured. A repeated measures ANOVA analysis was performed to determine differences between controls, MCI and AD patients. For each method we calculated effect sizes and the required sample sizes to detect one-year volume change between controls and MCI (N_{CTRL_MCI}) and between controls and AD (N_{CTRL_AD}). Finally, reproducibility of hippocampal atrophy rates was assessed using within-session rescans and expressed as an average distance measure D_Ave, which expresses the difference in atrophy rate, averaged over all subjects. The same D_Ave was used to determine the agreement between different methods.
Except for MALF, all methods detected a significant group difference between CTRL and AD, but none could find a significant difference between the CTRL and MCI. FreeSurfer and MIRTK required the lowest sample sizes (FreeSurfer: N_{CTRL_MCI} = 115, N_{CTRL_AD} = 17 with D_Ave = 3.26%; MIRTK: N_{CTRL_MCI} = 97, N_{CTRL_AD} = 11 with D_Ave = 3.76%), while ANTs was most reproducible (N_{CTRL_MCI} = 162, N_{CTRL_AD} = 37 with D_Ave = 1.06%), followed by Elastix (N_{CTRL_MCI} = 226, N_{CTRL_AD} = 15 with D_Ave = 1.78%) and NiftyReg (N_{CTRL_MCI} = 193, N_{CTRL_AD} = 14 with D_Ave = 2.11%). Manually measured hippocampal atrophy rates required largest sample sizes to detect volume change and were poorly reproduced (N_{CTRL_MCI} = 452, N_{CTRL_AD} = 87 with D_Ave = 12.39%). Atrophy rates of non-linear registration methods also agreed best with each other.
Non-linear registration methods were most consistent in determining hippocampal atrophy and because of their better reproducibility, methods, such as ANTs, Elastix and NiftyReg, are preferred for determining hippocampal atrophy rates on longitudinal MRI. Since performances of non-linear registration methods are well comparable, the preferred method would mostly depend on computational efficiency.
Brain atrophy in Alzheimer's Disease and aging
2016, Ageing Research Reviews
Thanks to its safety and accessibility, magnetic resonance imaging (MRI) is extensively used in clinical routine and research field, largely contributing to our understanding of the pathophysiology of neurodegenerative disorders such as Alzheimer’s disease (AD). This review aims to provide a comprehensive overview of the main findings in AD and normal aging over the past twenty years, focusing on the patterns of gray and white matter changes assessed in vivo using MRI. Major progresses in the field concern the segmentation of the hippocampus with novel manual and automatic segmentation approaches, which might soon enable to assess also hippocampal subfields. Advancements in quantification of hippocampal volumetry might pave the way to its broader use as outcome marker in AD clinical trials. Patterns of cortical atrophy have been shown to accurately track disease progression and seem promising in distinguishing among AD subtypes. Disease progression has also been associated with changes in white matter tracts. Recent studies have investigated two areas often overlooked in AD, such as the striatum and basal forebrain, reporting significant atrophy, although the impact of these changes on cognition is still unclear. Future integration of different MRI modalities may further advance the field by providing more powerful biomarkers of disease onset and progression.
Reproducibility of hippocampal atrophy rates measured with manual, FreeSurfer, AdaBoost, FSL/FIRST and the MAPS-HBSI methods in Alzheimer's disease
2016, Psychiatry Research - Neuroimaging
The purpose of this study is to assess the reproducibility of hippocampal atrophy rate measurements of commonly used fully-automated algorithms in Alzheimer disease (AD). The reproducibility of hippocampal atrophy rate for FSL/FIRST, AdaBoost, FreeSurfer, MAPS independently and MAPS combined with the boundary shift integral (MAPS-HBSI) were calculated. Back-to-back (BTB) 3D T1-weighted MPRAGE MRI from the Alzheimer's Disease Neuroimaging Initiative (ADNI1) study at baseline and year one were used. Analysis on 3 groups of subjects was performed – 562 subjects at 1.5 T, a 75 subject group that also had manual segmentation and 111 subjects at 3 T. A simple and novel statistical test based on the binomial distribution was used that handled outlying data points robustly. Median hippocampal atrophy rates were −1.1%/year for healthy controls, −3.0%/year for mildly cognitively impaired and −5.1%/year for AD subjects. The best reproducibility was observed for MAPS-HBSI (1.3%), while the other methods tested had reproducibilities at least 50% higher at 1.5 T and 3 T which was statistically significant. For a clinical trial, MAPS-HBSI should require less than half the subjects of the other methods tested. All methods had good accuracy versus manual segmentation. The MAPS-HBSI method has substantially better reproducibility than the other methods considered.
Adaptive registration of magnetic resonance images based on a viscous fluid model
2014, Computer Methods and Programs in Biomedicine
This paper develops a new viscous fluid registration algorithm that makes use of a closed incompressible viscous fluid model associated with mutual information. In our approach, we treat the image pixels as the fluid elements of a viscous fluid governed by the nonlinear Navier–Stokes partial differential equation (PDE) that varies in both temporal and spatial domains. We replace the pressure term with an image-based body force to guide the transformation that is weighted by the mutual information between the template and reference images. A computationally efficient algorithm with staggered grids is introduced to obtain stable solutions of this modified PDE for transformation. The registration process of updating the body force, the velocity and deformation fields is repeated until the mutual information reaches a prescribed threshold. We have evaluated this new algorithm in a number of synthetic and medical images. As consistent with the theory of the viscous fluid model, we found that our method faithfully transformed the template images into the reference images based on the intensity flow. Experimental results indicated that the proposed scheme achieved stable registrations and accurate transformations, which is of potential in large-scale medical image deformation applications.
Hippocampal volume change measurement: Quantitative assessment of the reproducibility of expert manual outlining and the automated methods FreeSurfer and FIRST
2014, NeuroImage
Citation Excerpt :
Furthermore, as fully automated methods may still be unsatisfactory and fully manual methods may be considered too costly with very large datasets, the performance of hybrid methods combining some expert input with automated processing should also be investigated. One example of this is the method previously described and validated by Crum et al. (2001) and van de Pol et al. (2007), which quantifies, using Jacobian integration of nonlinear image registration results, the total volumetric change in a region outlined by an expert on a single timepoint. By repeating the manual outlining of the hippocampi and registration procedures, van de Pol and colleagues found this hybrid method to have better reproducibility for atrophy rate measurement than the purely manual approach (Crum et al., 2001; van de Pol et al., 2007).
To measure hippocampal volume change in Alzheimer's disease (AD) or mild cognitive impairment (MCI), expert manual delineation is often used because of its supposed accuracy. It has been suggested that expert outlining yields poorer reproducibility as compared to automated methods, but this has not been investigated.
To determine the reproducibilities of expert manual outlining and two common automated methods for measuring hippocampal atrophy rates in healthy aging, MCI and AD.
From the Alzheimer's Disease Neuroimaging Initiative (ADNI), 80 subjects were selected: 20 patients with AD, 40 patients with mild cognitive impairment (MCI) and 20 healthy controls (HCs). Left and right hippocampal volume change between baseline and month-12 visit was assessed by using expert manual delineation, and by the automated software packages FreeSurfer (longitudinal processing stream) and FIRST. To assess reproducibility of the measured hippocampal volume change, both back-to-back (BTB) MPRAGE scans available for each visit were analyzed. Hippocampal volume change was expressed in μL, and as a percentage of baseline volume. Reproducibility of the 1-year hippocampal volume change was estimated from the BTB measurements by using linear mixed model to calculate the limits of agreement (LoA) of each method, reflecting its measurement uncertainty. Using the delta method, approximate p-values were calculated for the pairwise comparisons between methods. Statistical analyses were performed both with inclusion and exclusion of visibly incorrect segmentations.
Visibly incorrect automated segmentation in either one or both scans of a longitudinal scan pair occurred in 7.5% of the hippocampi for FreeSurfer and in 6.9% of the hippocampi for FIRST. After excluding these failed cases, reproducibility analysis for 1-year percentage volume change yielded LoA of ± 7.2% for FreeSurfer, ± 9.7% for expert manual delineation, and ± 10.0% for FIRST. Methods ranked the same for reproducibility of 1-year μL volume change, with LoA of ± 218 μL for FreeSurfer, ± 319 μL for expert manual delineation, and ± 333 μL for FIRST. Approximate p-values indicated that reproducibility was better for FreeSurfer than for manual or FIRST, and that manual and FIRST did not differ. Inclusion of failed automated segmentations led to worsening of reproducibility of both automated methods for 1-year raw and percentage volume change.
Quantitative reproducibility values of 1-year microliter and percentage hippocampal volume change were roughly similar between expert manual outlining, FIRST and FreeSurfer, but FreeSurfer reproducibility was statistically significantly superior to both manual outlining and FIRST after exclusion of failed segmentations.
Standardization of MRI and Amyloid Imaging
2013, Global Clinical Trials for Alzheimer's Disease: Design, Implementation, and Standardization
In-vivo imaging markers of neuronal changes related to Alzheimer’s disease (AD) are ideally suited to be employed as diagnostic markers for early and differential diagnosis of AD as well as for the assessment of neurobiological effects of medical treatments in clinical trials. Novel molecular imaging techniques enable in-vivo detection of cerebral amyloid pathology, whereas magnetic resonance imaging (MRI)-based techniques, such as volumetric MRI and diffusion tensor imaging (DTI), provide structural lesion markers that allow tracking disease progression from preclinical through predementia to clinically manifest stages of AD. However, a widespread clinical use of these imaging biomarkers is hampered by considerable multi-centric variability related to differences in scanner hardware and acquisition protocols, but also by the lack of internationally agreed upon standards for analytic design and employed quantitative metrics.
Several strategies for reducing multicenter variability in imaging measures have been proposed, including homogenization of the acquisition settings across scanner platforms, stringent quality assurance procedures, and artifact removal by means of post-acquisition image processing techniques. In addition, selection of appropriate statistical models to account for remaining multicenter variability in the data can further improve the accuracy and reproducibility of study results. The first projects for international standardization of image analysis methods and derived quantitative metrics have emerged recently for volumetric MRI measures. In contrast, the standardization and establishment of DTI-derived measures within a multicenter context are less well developed. Although molecular imaging techniques are already widely used in multicenter settings, sources of variability across sites and appropriate methods to reduce multicenter effects are still not explored in detail. Comparability of neuroimaging measures as AD biomarkers in worldwide clinical settings will finally depend on the establishment of internationally agreed upon standards for image acquisition, quality assurance, and employed quantitative metrics.

View all citing articles on Scopus

View full text

Improved reliability of hippocampal atrophy rate measurement in mild cognitive impairment using fluid registration

Abstract

Introduction

Section snippets

Subjects

Results

Discussion

Acknowledgments

NeuroImage

Neurobiol. Aging

NeuroImage

Neurobiol. Aging

Comput. Methods Programs Biomed.

NeuroImage

NeuroImage

Lancet

Acta Neuropathol. (Berl.)

Neurology

J. Comput. Assist. Tomogr.

Epilepsia

Neurology