10.1177/1051228404266264
Journal of Neuroimaging Vol 14 No 3(Supplement) July 2004
Pelletier et al: Brain Atrophy in MS
Measurement of Whole-Brain
Atrophy in Multiple Sclerosis
Daniel Pelletier, MD
Kathleen Garrison, MSc
Roland Henry, PhD
ABSTRACT
Brain atrophy reflects the net result of irreversible and destruc-
tive pathological processes in multiple sclerosis (MS). The gross
morphological changes can be accurately quantified using
standard magnetic resonance imaging (MRI) acquisitions and
various image analysis tools. The current methods used to
assess whole-brain atrophy in patients with MS can be classi-
fied into 2 groups based on their reliance on segmentation and
registration. Segmentation-based methods employed to mea-
sure whole-brain atrophy in MS include the brain parenchymal
fraction, the index of brain atrophy, the whole-brain ratio, the
brain to intracranial capacity ratio, fuzzy connectedness/
Udupa’s method, 3DVIEWNIX, the Alfano method, and SIENAX.
Current registration-based methods used to measure whole-
brain atrophy in MS include the brain boundary shift integral,
SIENA, statistical parametric mapping, template-driven seg-
mentation, and voxel-based morphometry. Most of the meth-
ods presented here are sensitive to subtle changes in brain
structures and have been successfully applied to MS as mea-
sures of whole-brain atrophy. Yet comparative studies of these
methods are limited and are complicated by the lack of a gold
standard for image acquisition, a segmentation algorithm, an
image analysis method, or a reproducibility measure. Overall,
the measure of whole-brain atrophy from MRI contributes to an
appreciation of the dynamics of MS pathology and its relation-
ship to the clinical course of MS. Determination of the relative
reproducibility, precision, sensitivity, and validity of these meth-
ods will promote the use of whole-brain atrophy measures as
components of comprehensive MRI-based outcome
assessment in MS clinical trials.
Key words: Multiple sclerosis, brain atrophy, magnetic reso-
nance imaging (MRI) techniques.
Pelletier D, Garrison K, Henry R.
Measurement of whole-brain
atrophy in multiple sclerosis.
J Neuroimaging 2004;14:11S-19S.
DOI: 10.1177/1051228404266264
Brain atrophy reflects the net result of irreversible and
destructive pathological processes in multiple sclerosis
(MS). Axonal damage and loss, chronic demyelination,
and gliosis all contribute to a reduction in brain
parenchymal tissue volume and a corresponding expan-
sion of cerebrospinal fluid (CSF) spaces. These gross
changes in morphology can be accurately quantified
using standard magnetic resonance imaging (MRI) acqui-
sitions and various image analysis tools. Because brain
volume changes over time are small, accurate calculation
of whole-brain atrophy requires a highly reproducible
measurement, and MRI can detect subtle changes in
brain structure. Conventional MRI sequences can be
used to estimate brain tissue volumes and areas, which are
then used to quantify whole-brain atrophy in cross-
sectional or longitudinal value comparisons. Measure-
ment of whole-brain atrophy using MRI has emerged as
an objective and reliable marker of disease progression in
patients with MS and as a potential outcome measure in
clinical trials of treatments for MS. Although the optimal
method for quantifying whole-brain atrophy in MS has
not yet been established, this article reviews the current
approaches and considerations in the measurement of
whole-brain atrophy. A summary of the specific methods
discussed in this article is provided in Table 1.
Methodological Concepts
The quantification of brain tissue volume is the basic con-
cept underlying all methods for the measurement of brain
atrophy from MRI. Although several different methods
are available, a few fundamental concepts and techniques
serve as the defining characteristics of atrophy assessment
methods. These concepts are described below and in-
clude normalization, automation, and registration.
Normalization
An important distinction among volumetric measure-
ments of whole-brain atrophy is whether brain size is re-
ported as actual volume or as normalized volume. In
Copyright © 2004 by the American Society of Neuroimaging 11S
Received March 1, 2004. Accepted for publication
March 30, 2004.
From the Multiple Sclerosis Center, University of Califor-
nia San Francisco.
Address correspondence to Dr Daniel Pelletier, UCSF
Multiple Sclerosis Center, 350 Parnassus Avenue, Suite
908, San Francisco, CA 94117. E-mail: danp@
itsa.ucsf.edu.