[Frontiers in Bioscience 9, 647-664, January 1, 2004]


Robert Zivadinov 1-3, and Rohit Bakshi 1-4

1 Buffalo Neuroimaging Analysis Center, 2 The Jacobs Neurological Institute, 3 Department of Neurology, School of Medicine and Biomedical Sciences, University at Buffalo, State University of New York, and 4 Physicians Imaging Centers, Buffalo, NY


Figure 1. Brain atrophy in multiple sclerosis shown on MRI scans. Non-contrast T1-weighted images are shown of normal controls in the fifth decade (a-b) compared with age-matched patients with MS (c-d). Note thinning and decreased volume of corpus callosum and posterior fossa structures in MS (c). Note prominence of cortical sulci (suggesting cortical atrophy) and ventriculomegaly (indicating central atrophy) in MS (d). Using this type of visual analysis, ordinal rating systems can semiquantitatively measure atrophy (113).

Figure 2. Semiautomated method of determining brain parenchymal fraction, a normalized measure of whole brain atrophy in MS, from the Buffalo Neuroimaging Analysis Center (adapted in part from ref. 37). (a-c): A single axial slice, spin-echo T1-weighted noncontrast, from an MS patient showing the raw image (a), after masking (removal) of extracranial tissue has been performed to isolating the outer brain contour (b), and after thresholding to separate the intracranial volume into brain parenchyma (black pixels) and CSF (white pixels) (c). (d-f): Shows that the same algorithm is applied to all axial images from the mid-cerebellum to the vertex. The image surrounded by the large white square (f) is used to identify normal appearing white matter on which a region-of-interest is placed in the normal appearing white matter (b, small white box) to determine the threshold for parenchyma vs. CSF segmentation (37). The BPF is the ratio of the brain parenchymal volume to the intracranial volume (volume within the surface contour including the subarachnoid space) and is thus a normalized measure. With recent software improvements, the total analysis time per patient is 10 minutes.

Figure 3. Whole brain atrophy in MS as measured by brain parenchymal fraction (BPF) from the Buffalo Neuroimaging Analysis Center (adapted in part from ref. 37). BPF is the ratio of the brain parenchymal to intracranial volume. Representative mid-ventricular axial MRIs are shown from non-contrast T1-weighted images of age-matched individuals in the fourth decade: (a) - normal volunteer; (b) -relapsing-remitting MS, mild-moderate physical disability, disease duration of 6 years; (c) -secondary progressive MS, moderate-severe disability, disease duration 12 years. Figures (d) and (e) depict whole brain atrophy in patients with MS vs. normal individuals (37). Bargraphs (d) show mean BPF and std. error of mean in MS (n=50, MS-all) and age-/sex-matched controls (n=17, NL). BPF was lower in MS (p< .001), indicating whole brain atrophy in MS. Both MS subgroups had lower BPF as compared to NL; BPF was lower in relapsing-remitting (RR) MS (n=40, p= .005. vs. NL) and in secondary progressive (SP) MS (n=10, p=.0005). Asterisks indicate a significant difference compared to NL. A scatter plot (e) shows the relationship between whole brain atrophy and physical disability in 78 patients with MS; brain parenchymal fraction was inversely related to overall physical disability (Expanded Disability Status Scale) score (r=-.391, p=.0006).

Figure 4. Regional brain atrophy measurement of the frontal lobes in a patient with relapsing-remitting multiple sclerosis. (a), (b) and (c) shows masked T1-weighted images of the frontal lobes segmented at various levels and obtained by semiautomated iterative morphologic outlining of the external cerebrospinal fluid (CSF) spaces; the next six images show the result of segmentation: (d), (e) and (f) are brain parenchyma-only images and (g), (h) and (i) are CSF-only images.

Figure 5. Quantitative regional atrophy analysis techniques in MS from the Buffalo Neuroimaging Analysis Center (adapted in part from refs. 17 and 18). (a): The bicaudate ratio (BCR) (17) or intercaudate nucleus ratio (19) is the minimum intercaudate distance (solid line) divided by brain width along the same line (dashed line). BCR is measured in an axial slice where the heads of the caudate nuclei are most visible and closest together. BCR is elevated in patients with MS, consistent with subcortical atrophy, and is independently associated with cognitive dysfunction (17). (b): A method (8,9) of measuring third ventricle width (3VW), a marker of central atrophy, is shown in a patient with MS. This involves drawing a line through the long axis of the third ventricle, parallel to the inter-hemispheric fissure. The width is measured by drawing a second line perpendicular to the first at its midpoint (arrow). Greater physical disability increments over one and two years in relapsing-remitting patients with MS have been associated with increasing 3VW (9). (c-d): Parcellation of the caudate nuclei in MS (18) involving manually tracing regions of interest delineating the caudate nuclei bilaterally in each slice in which they are visible on high resolution gradient echo coronal T1-weighted images (c). The region of interest files are then reconstructed, normalized, and compared with age-matched normal controls (NL). (d): Superior views of overlaid caudate nuclei from five representative MS (dark gray) and NL subjects (light gray) reveal caudate atrophy in MS; volumes were nearly 20% less than NL (18).

Figure 6. Progressive brain atrophy in MS over 18 months shown on representative MRI scans. A 22 year-old woman with relapsing-remitting MS and a disease duration of two years received MRI of the brain (top row). She had mild physical disability on the Expanded Disability Status Scale (EDSS) of 1.0. Brain parenchymal fraction (BPF), performed at the Buffalo Neuroimaging Analysis Center, was 0.88, which is normal for age. The patient experienced frequent relapses and disease progression during the next several months. Physical disability increased to a moderate level with an EDSS score of 4.0. A follow-up MRI was performed (remote from corticosteroid use) 18 months after the initial MRI. Note the marked enlargement of the cortical sulci, sylvian fissures, and third and lateral ventricles. BPF decreased by 4.5% to 0.84, which is ~5 standard deviations below normal.

Figure 7. Representative tomographic MRI slice showing segmentation into grey and white matter obtained using Statistical parametrical mapping (SPM99) at the Buffalo Neuroimaging Analysis Center. Shown are (a) the source image (high-resolution T1-weighted gradient echo), (b) after masking and segmentation into brain parenchyma (black) and CSF (white), (c) segmentation into white matter (white areas), and (d) segmentation into grey matter (white areas).

Figure 8. Output of 3 methods of determining brain parenchymal fraction (BPF) in a 32-year old man with relapsing-remitting MS. BPF was quantified on conventional spin-echo T1-weighted axial sequences by 3 software approaches. The Buffalo (37) and Trieste (41) methods were semiautomated, while SIENAX (52) was fully automated. Axial T1-weighted images (far left), scalped- (left middle), brain parenchyma- (right middle) and CSF- only images (far right), are presented. There are clear differences in how semiautomated (Buffalo and Trieste) and automated (SIENAX) algorithms segmented the images into brain parenchyma- and CSF-only images, translating to differences in clinical correlation and precision (66).

Figure 9. MRI T2 hypointensity (T2H) in the gray matter and brain atrophy in MS. Hypointensity on T2-weighted images has been described in the gray matter of patients with MS and is related to physical disability, clinical course, MRI lesion load, and brain atrophy (8,91,92). The finding most likely represents pathologic iron deposition. Conventional spin-echo T2-weighted images are shown of a normal volunteer (a) and an age-matched patient with secondary progressive MS early in the fourth decade of life (b). In the latter, note the marked hypointensity of the deep gray matter nuclei, including the thalamus and putamen. A natural history longitudinal study (93) of 68 relapsing-remitting patients revealed that T2 hypointensity (T2H) in the caudate, putamen, globus pallidus, red nucleus, and thalamus correlated with baseline brain parenchymal fraction (BPF) (r=.19-.39, p=.001-.029). Baseline T2H in the thalamus, putamen, and globus pallidus predicted 2-year BPF change (r=.26-.33, p=.006-.032) and T2H was chosen in regression modeling as the best predictor of BPF change after accounting for all MRI lesion variables (c). Thus, gray matter T2H predicts progression of brain atrophy suggesting a relationship between iron deposition and tissue destruction in MS.