2.1 Demographics
Data used in this study were obtained from Alzheimer’s Disease Neuroimaging Initiative (ADNI) (adni.loni.usc.edu). Among ADNI database, we analyzed the subjects who took both MRI and PET (amyloid, AV 45 and tau, AV 1451): 78 cognitive normal (CN), 50 EMCI, 34 LMCI, and 39 AD. Subjects were sampled with following criteria: age around 60 to 90-year-old, education year 12 to 20, and gender match within group. To assess AD continuum, amyloid negative CN and amyloid positive EMCI, LMCI, and AD subjects were selected. EMCI group was subdivided into 38 dementia non-converter (stable EMCI) and 12 converter group to assess changes with disease progression. A total of 201 subjects’ T1 and DTI images were gathered from ADNI. To increase sample size, multi-center approach was used as discussed in [13]. The amyloid positivity of the subjects was determined by whole brain PET AV45 standardized uptake value ratio (SUVR) with 1.11 cut-off. Table 1 shows the demographics of the subjects used in this study; note that subjects who underwent AV1451 tau PET imaging were 44 in CN, 9 in EMCI, 5 in LMCI, and 3 in AD. Additional 28 CN subjects who showed amyloid positive were gathered to identify earliest AD pathological changes as presented in Supp. Table 1.
2.2 Image processing
T1 weighted images were processed with Freesurfer package v6.0 (http://surfer.nmr.mgh.harvard.edu) as previously reported in [13]. Cortical thickness (CTh) maps were registered to Freesurfer average sphere through spherical registration for group comparison. DTI and PET images were registered with their respect to T1 images using boundary-based algorithm for further process. DTI images were processed using FSL package as followed: eddy current correction, rotate gradient vectors from the results of eddy correction, and tensor fitting to produce mean diffusivity map and primary eigenvector map. DTI metrics were further processed to avoid partial volume effect using Koo et al [27]. PET images were partial volume corrected using mri_gtmpvc which is built in Freesurfer package. PET images were normalized by mean signal from whole cerebellum and converted to SUVR for amyloid and tau PET, AV45 and AV1451 respectively. Then images were boundary-based registered to corresponding T1 structural images. To avoid any partial volume effects, the center parts of the cortical column were sampled for surface analysis. Lastly, CTh was smoothed with 10 mm while other modalities were smoothed with 15 mm full width half maximum Gaussian kernel. Fig. 1 shows the overall scheme of the process.
2.3 Calculation of radiality
A surface normal vector was obtained from individual gray matter surface to define cortical orientation. Freesurfer represent the surface in triangular meshes, and surface normal vector can be computed using cross-product between edges. Vertex-wise dot product between primary diffusion direction, primary eigenvector of diffusion tensor, and the surface normal vector was quantified as a radiality index: r: where v represents surface normal vector and e1 represents primary diffusion direction [22].
See formula 1 in the supplementary files.
It ranges from 0 to 1, where r = 0 indicates tangential diffusion and r = 1 indicates radial diffusion to cortex. Subject’s principal eigenvector map was projected onto the individual surface reconstruction to calculate vertex-wise radiality as discussed in [22].
2.4 Cut-off analysis
To further test feasibility of radiality as AD biomarker, we performed cut-off analysis using receiver operating characteristics graphs to distinguish CN with other AD stages as shown in Supp. Table 2. The feature used was the mean radiality within cluster that obtained from CN vs EMCI group comparison. With varying cut-off, we sought to find the cost-effective point where it minimizes the difference between sensitivity and specificity [23].
2.5 Statistical analysis
We first compared the differences between groups for radiality, CTh, MD, AV45 and AV1451 with a general linear model, which is available in Freesurfer. The results were cluster-wise corrected for family-wise error (FWE) corrected p-value < 0.05
To test the associations between radiality and other neuroimage biomarkers, we calculated set of vertex-wise partial correlations with the radiality as the dependent variable and CTh, MD, AV45, and AV1451 as the independent variable. Age, gender, year of education, and MRI center were set as covariates of cluster analyses. Permutation test was applied to resolve multiple comparisons problem through a Monte Carlo simulation with 10,000 repeats, which is built-in function of Freesurfer.
To test the linear relationship between radiality and other neuroimage biomarkers, we quantified mean metrics within AD specific ROIs. ROIs include entorhinal, fusiform, insula, inferior, middle, and superior temporal cortex. Mean metrics within ROIs were plotted in a box and whisker plots and presented in Fig. 5(e) and Fig. 6. Significance between groups was tested with one-way ANOVA.