Loredana Storelli1, Maria A. Rocca1,2, Elisabetta Pagani1, Wim Van Hecke3, Mark A. Horsfield4, Nicola De Stefano5, Alex Rovira6, Jaume Sastre-Garriga7, Jacqueline Palace8, Diana Sima3, Dirk Smeets3, and Massimo Filippi1,2
1Neuroimaging Research Unit, INSPE, Division of Neuroscience, San Raffaele Scientific Institute, Vita-Salute San Raffaele University, Milan, Italy, 2Department of Neurology, San Raffaele Scientific Institute, Vita-Salute San Raffaele University, Milan, Italy, 3Research and Development for Icometrix, KU Leuven, Leuven, Belgium, 4Xinapse Systems, Colchester, United Kingdom, 5Department of Medicine, Surgery and Neuroscience, University of Siena, Siena, Italy, 6Section of Neuroradiology, Department of Radiology, Hospital Universitari Vall d’Hebron, Barcelona, Barcelona, Spain, 7Unitat de Neuroimmunologia Clinica, CEM-Cat, Hospital Universitari Vall d’Hebron, Barcelona, Spain, 8Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom
Synopsis
We
compared different methods for whole-brain and grey matter (GM) atrophy
estimation (ANTs v1.9, CIVET v2.1, FSL-SIENA(X) v5.0.1, Icometrix-MSmetrix v1.7,
and SPM v12) in multiple sclerosis (MS). The accuracy and precision were
evaluated for cross-sectional and longitudinal whole-brain and GM atrophy
measures. All software showed high accuracy and comparable repeatability for cross-sectional
measures. However, since there was poor reproducibility and high variability in
cross-sectional and longitudinal atrophy measures, changes of MR scanner should
be avoided. This study may help in the selection of a suitable pipeline,
depending on the requirements of the application (research center, clinical setting
or clinical trial).
Introduction
Neurodegeneration is a clinically relevant pathological
hallmark of multiple sclerosis (MS).1,2 The quantification of volume
loss (atrophy) of the brain and of the grey matter (GM) from magnetic
resonance imaging (MRI) is widely accepted as an in vivo biomarker of neurodegeneration in MS and many other
neurological conditions.3-6 Halting neurodegeneration and promoting
neuroprotection is a prime goal of current therapeutic strategies.7,8
Several software tools are
available to measure atrophy from MRI, although none is routinely used in
clinical practice.9-12 The aims of this study were the assessment
and comparison of different automatic or semi-automatic methods (ANTs version
1.9, CIVET version 2.1, FSL-SIENAX/SIENA 5.0.1, Icometrix-MSmetrix 1.7, and SPM
version 12) currently available for whole-brain and grey matter (GM) atrophy
quantification in MS (both cross-sectionally and longitudinally), and an appraisal
of the feasibility of moving them into the clinical setting.Methods
The dataset arranged for this study consisted of 3DT1 and 3D-T2 FLAIR MRI sequences of
simulated data of MS brains, longitudinal data from healthy controls (HC), test
and retest MRI of MS patients acquired at different MR scanner field strengths
and manufacturers, and longitudinal data from MS patients (1 year of
follow-up). A MRI simulator was developed
for the creation of the simulated data. The digital brain phantoms with mild
and severe lesion load (respectively 0.42 and 10.1 ml), tissue MR parameters
and Intensity non-uniformity (INU) fields available from BrainWeb were used.13 Standard 1.5T parameters were included into the Bloch
equation to obtain T1-weighted and FLAIR sequences:
S(x,y,z)=ρ(x,y,z)|1-2exp(-TI/T1(x,y,z) ) |*[1-exp(-TR/T1(x,y,z) ) ]*exp(-TE/(T2(x,y,z)));
Cross-sectional and longitudinal whole-brain and GM
atrophy estimation were tested for each software package. For ANTs and SPM,
longitudinal pipelines were implemented using their own tools, according to
Jacobian integration method.14 For the validation, we used
test-retest MRI of MS patients acquired using the same scanner to evaluate
repeatability, while test-retest scans acquired using different scanner
manufacturers and different MR field strengths were used to assess the
reproducibility of brain and GM volume measures. Simulated data of MS patients and
longitudinal dataset from HC were used to assess the accuracy of respectively whole-brain
and GM cross-sectional and longitudinal atrophy measures. Longitudinal data
from MS patients (1 year of follow-up) were used to assess the agreement
between atrophy results for the different methods, using the Intra-class
correlation coefficient (ICC). Moreover,
the main steps that are common to all processing pipelines were qualitatively
and quantitatively evaluated [brain and GM segmentation, image registration, white
matter (WM) lesion filling].
Results
High values of accuracy (0.87-0.97) on simulated dataset for
whole-brain and GM volume measures were found, with the lowest values for
MSmetrix (0.87-0.88). ANTs showed the smallest mean error in estimating percentage
of whole-brain volume changes on HC (0.02%) with a coefficient of variation
(CoV) of 0.5% (Figure 1). SPM showed the smallest mean error (0.07%) and CoV
(0.08%) in estimating percentage of GM volume changes. Good repeatability was
found on average for whole-brain and GM volume measures for all software
(Figure 2), but a poor reproducibility between the results from different MR
field strengths (Figure 3) and manufacturers (Figure 4) was found. Regarding
the sensitivity to detect atrophy changes in MS patients with one year of
follow-up, a significant ICC was found between SIENA and SPM whole-brain
longitudinal atrophy results, while no significant agreement was found between
GM atrophy results for the different methods (Figure 5). From image registration
assessment, comparable values of normalized mutual information (NMI) were found
for the different pipelines (NMI>1), for image registration between subject
to atlas and between two time points of the same subject. The WM lesion filling
technique mainly affected longitudinal atrophy results for ANTs, MSmetrix and
SPM packages.Discussion
All pipelines showed comparable
repeatability of whole-brain and GM volume quantification when the input data were
carefully controlled (consistent patient positioning and pulse sequence in a
single scanner). However, changes
of MR scanner should be avoided and an improved reproducibility is required to
all pipelines. Moreover, the coefficients of
variation estimated from repeated measures showed that the variability was too
high to allow individualized patient studies and clinical application.Conclusion
This study may help in the selection of a suitable
pipeline from among those available, depending on the requirements of the
analysis framework (research center, clinical setting or clinical trial), and
whether the goal is high accuracy and repeatability or high reproducibility. These
results may also be helpful to directing further improvements to atrophy
processing pipelines for future clinical use in MS.Acknowledgements
No acknowledgement found.References
1.
Evangelou N, Esiri MM, Smith S, et al. Quantitative pathological evidence for
axonal loss in normal appearing white matter in multiple sclerosis. Ann Neurol 2000;
47: 391-395.
2. Popescu BF, Pirko I,
Lucchinetti CF. Pathology of multiple sclerosis: where do we stand? Continuum (Minneap Minn) 2013; 19: 901-921.
3. Filippi M, Preziosa
P, Copetti M, et al. Gray matter damage predicts the
accumulation of disability 13 years later in MS. Neurology 2013; 81: 1759-1767.
4.
Miller DH, Barkhof F, Frank JA, et al. Measurement of atrophy in multiple
sclerosis: pathological basis, methodological aspects and clinical relevance. Brain 2002; 125: 1676-1695.
5. Pravatà E, Rocca MA, Valsasina P, et al. Gray matter
trophism, cognitive impairment, and depression in patients with multiple
sclerosis. Mult Scler 2017 doi: 10.1177/1352458517692886.
6.
van Munster CE, Jonkman LE, Weinstein HC, et al. Gray matter damage in multiple
sclerosis: Impact on clinical symptoms. Neuroscience 2015; 303: 446-461.
7.
Sormani MP, Arnold DL, De Stefano N. Treatment effect on brain atrophy
correlates with treatment effect on disability in multiple sclerosis. Ann
Neurol 2014; 75: 43-49.
8.
Branger P, Parienti JJ, Sormani MP, Defer G. The Effect of Disease-Modifying
Drugs on Brain Atrophy in Relapsing-Remitting Multiple Sclerosis: A
Meta-Analysis. PLoS One 2016; 11: e0149685.
9.
Avants BB, Tustison NJ, Wu J, et al. An open source multivariate framework for
n-tissue segmentation with evaluation on public data. Neuroinformatics 2011; 9:
381-400.
10.
Smith SM, Zhang Y, Jenkinson M, et al. Accurate, robust, and automated
longitudinal and cross-sectional brain change analysis. Neuroimage 2002; 17:
479-489.
11.
Smeets D, Ribbens A, Sima DM, et al. Reliable measurements of brain atrophy in
individual patients with multiple sclerosis. Brain Behav 2016; 6: e00518.
12.
Zijdenbos AP, Forghani R, Evans AC. Automatic "pipeline" analysis of
3-D MRI data for clinical trials: application to multiple sclerosis. IEEE Trans
Med Imaging 2002; 21: 1280-1291.
13. BrainWeb:
Simulated Brain Database. www.bic.mni.mcgill.ca/brainweb/.
14. Nakamura
K, Guizard N, Fonov VS, et al. Jacobian integration method increases the
statistical power to measure gray matter atrophy in multiple sclerosis.
Neuroimage Clin 2014; 4: 10-17.