Change Detection and Quantification
in Multiple Sclerosis
D. Louis Collins
Sept 26, 2004
Montreal Neurological
Institute/Hospital
McGill University
Multiple Sclerosis
• Motivations
• Volume change
– Global (BICCR)
– Regional (GM, ventricels, lobes)
– Local (around lesions)
• Clinical trial
– BICCR results
– VBM results
• Deformation modeling
– Where and When?
Motivation
• Clinical surrogates of disease burden in MS are
highly variable (EDSS, MSFC)
• MRI shows lesions in vivo
T1-w
PD
T2-w
MTR
Gado
Motivation
• Clinical surrogates of disease burden in MS are
highly variable (EDSS, MSFC)
• MRI shows lesions in vivo
• MRI = 10 * clinical activity
MRI activity
QuickTime™ and a
YUV420 codec decompressor
are needed to see this picture.
MRI shows brain atrophy in MS
normal
MS
Motivation
• Clinical surrogates of disease burden in MS are
highly variable (EDSS, MSFC)
• MRI shows lesions in vivo
• MRI = 10 * clinical activity
> MRI-based surrogates of disease burden
MRI-based surrogates
• T2 and Gado-based lesion metrics
– have shown treatment effects
– are weakly correlated with disability
• CNS atrophy
– associated with neuronal/axonal loss
– associated with irreversible neurological impairment
– strong correlations with disability
CNS atrophy may be a better surrogate
Methodological Requirements
• Reproducible
• Sensitive to change
• Accurate
• Practical
Data acquisition issues
• Resolution requirements
– Thin slices to reduce partial volume effects
– Contiguous acquisitions (no slice gap)
– Prefer 3D acquisitions over 2D
• Contrast
– T1 with or w/o T2/PD
• Time constraints
– Short acquisition to minimize motion artifacts
threshold, followed
by erosion + dilation
Tissue
classification
PDw MRI
T2w MRI
classified data
ICC mask
CSF
LESION
WM
GM
LESION
WM
GM
BICCR
+
+
+
+
+
=
BICCR: Brain to IntraCranial Capacity Ratio
Measuring Changes in Brain Volume
Atrophy
• Scan-rescan COV of BPF, BICCR = 0.2%
• Smallest detectable change ~0.5%
BICCR by Age: Normal Controls
15
20
25
30
35
40
AGE
0.75
0.80
0.85
0.90
0.95
BICCRmen
women