Premium
IC–P–021: Multispectral MRI analysis in comprehensive assessment of the aging brain
Author(s) -
Anderlik Andrea,
Ystad Martin,
Bergmann Ørjan,
Lundervold Astri J.,
Geitung Jonn-Terje,
Reinvang Ivar,
Lundervold Arvid
Publication year - 2006
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2006.05.2226
Subject(s) - white matter , fractional anisotropy , magnetic resonance imaging , corpus callosum , diffusion mri , voxel , artificial intelligence , computer science , nuclear medicine , medicine , pathology , radiology
Background: Multiparametric magnetic resonance imaging enables local assessment of tissue “signatures” in the aging brain. By proper alignment of data we can for each voxel, or region of interest (ROI) i obtain a pattern vector i ( i1, . . . , ip), where ij expresses a local tissue property. Such pattern vectors can jointly be combined with test results from genotyping and cognitive evaluation, and thereby give important and differentiating information in normal aging, mild cognitive impairment and Alzheimers disease. Objective(s): The present work is a feasibility study of combining advanced multispectral imaging methods and genetic and behavioral data in the study of normal aging. We apply our methods to the first subject in a large series (N 109) of recently obtained examinations. Methods: Our MR imaging protocol comprises anatomical dual-volume acquisitions for tissue segmentation and morphometric analysis (FreeSurfer), and dual-echo PD/T2 weighted acquisitions followed by diffusion tensor imaging for white matter analysis (DTIStudio). For intrasubject image registration we utilized geometric information in the DICOM headers to obtain proper spatial transformations and reslicing. To complement our pattern vectors, we have genotypes for APOE [alleles: 2 3 4], CHRNA4 [T C], BDNF [val met], and results from extensive neuropsychological testing, e.g. WASI (IQ), CVLT (verbal memory), and PASAT (attention). Results: In the present case, where we have computed mean fractional anisotropy (FA) and mean ADC [10 m/s] in the genu corpus callosum (‘gcc’) as well as total brain volume and volumes of the left and right hippocampus [ml], the following pattern vector was obtained: (subj_id, sex, age, edu, IQtot, CVLTtot, PASATsum, . . . , APOE, CHRNA4, BDNF, . . . , FA_gcc, ADC_gcc, brain_vol, hippo_left, hippo_right) (501, F, 53, 14, 122, 54, 41, . . . , 3/ 3, C/C, val/val, . . . , 0.71, 0.73, 1325, 3.7, 3.9). Conclusions: By careful planning of the MR imaging protocol and utilization of image registration techniques we can obtain voxelbased and ROI-based pattern vectors that express a rich variety of tissue information (cf. Fig). Joining these vectors with genotypes and results from neuropsychological phenotyping, multivariate statistical analysis can be used to reveal new insight into normal and abnormal aging of the brain.