Gunter, Jeffrey L.

AbstractBackgroundWe have previously shown that 3D T1‐weighted and 3D T2‐FLAIR brain MRI can potentially be re‐identified using face recognition, but this has not been tested for other MRI sequences commonly used in imaging studies of aging and dementia. Here we compare match rates of automated face recognition to re‐identify 3D T1‐weighted, 3D T2‐FLAIR, 3D T2‐weighted, 2D Axial T2‐FLAIR, multi‐echo GRE (T2*), diffusion MRI, functional MRI, ASL, and T2* GRE images.MethodWe used a dataset of 182 Mayo Clinic participants each with MRI and face photographs. MRIs used Siemens and GE vendor‐product sequences similar to ADNI3 and ADNI4 protocols. We created face reconstructions (“basic”, with minimal preprocessing, and “advanced” that automatically filled missing face regions using a template) from each MRI sequence, and we trained a Microsoft Azure PersonGroup face recognition classifier to recognize these MRI‐based faces. We then queried the classifier with five face photographs for each participant, and we recorded how often the classifier chose the correct MRI as the top‐ranked match. For image types with match rates > = 10%, we then repeated the experiment after de‐facing each MRI with mri_reface, our leading software that replaces faces in each image (“re‐faces”) with an average face.ResultResults are presented in Table 1. 3D T2‐FLAIR, 3D T1‐weighted, and 3D T2‐weighted images were matched at 96‐98%, and mri_reface reduced all to 8%. Axial T2‐FLAIR and ME‐GRE were matched at 32‐45%, and mri_reface reduced both to 1%. T2*‐GRE matched at 10%, and mri_reface reduced this to 1%. ASL matched at 8%, while dMRI and fMRI matched at only 0‐2%.ConclusionAmong current‐generation vendor‐product sequences, we recommend that de‐facing be applied before research data sharing of T1‐weighted, T2‐weighted, T2‐FLAIR, T2*, and ME‐GRE images. Minimal match rates for current‐generation product fMRI, ASL, and fMRI images suggest that they present low risk of re‐identification and can be shared without de‐facing, but this conclusion should be re‐evaluated if they are acquired without fat‐suppression, with a larger field of view, or if newer developments reduce the current levels of artifacts and distortion around the face.

AbstractBackgroundPredictive models of tau‐PET accumulation have largely relied on the functional connectome to estimate its spatial patterns in Alzheimer’s disease (AD). Patterns of global functional organization represent a promising avenue as they encapsulate many biological properties relevant to neurodevelopment, cognition and degeneration. We compared approaches based on local connectivity (from task‐free fMRI) and global organization (from FDG‐PET) to predict regional tau‐PET across AD phenotypes.MethodWe included 430 cognitively unimpaired (CU) individuals (214 A+T‐, 216 A+T+) and 62 amnestic mild cognitive impairment (aMCI), 64 amnestic AD, 21 dysexecutive AD (dAD), 41 logopenic progressive aphasia (LPA) and 49 posterior cortical atrophy (PCA) patients (all A+T+) from the Alzheimer’s Disease Research Center, Mayo Clinic Study of Aging, and Neurodegenerative Research Groups (referred to as “Mayo cohort”). The connectivity approach consisted of assessing the relationship between a region’s tau level and its functional connectivity to the region of highest tau (epicenter). The global organization approach consisted of 1) performing a principal component analysis on 3000 FDG‐PET images that did not overlap with the Mayo cohort, 2) projecting FDG‐PET images of the Mayo cohort onto the low‐dimensional space generated in Step 1 and extracting corresponding weighted values across the first 100 components, 3) using these weighted values as predictors of regional tau‐PET in a linear model and determining optimal parameters based on predicted R2 values, and 4) using the parameters from the model built in Step 3 to predict tau‐PET in the ADNI cohort (n = 86) (Fig.2).ResultThe connectivity‐based approach yielded adjusted R2 values ranging from 0.19‐0.45 across diagnostic groups (Fig.1). The global organization approach yielded averaged adjusted R2 and predicted R2 values of 0.71 and 0.63, respectively, in the Mayo cohort. Out‐of‐sample median R2 in the ADNI cohort was 0.41 (range 0.10‐0.60) (Fig.3).ConclusionAn unbiased approach based on the global functional organization of the brain and using an imaging modality widely used in clinical settings outperformed a connectome‐based one to predict regional tau‐PET across the AD phenotypic spectrum. Patterns of tau accumulation may better relate to the large‐scale physiology of the brain underlying specific mental functions rather than local connectivity factors.

AbstractBackgroundAutomated face recognition can potentially re‐identify de‐identified research brain MRI, PET, and CT, and direct differences of brain measurements from original and de‐faced images are statistically significant but very minor, i.e. below scan‐rescan differences. Their effects on analyses correlating brain imaging biomarkers with clinical variables remain unknown.MethodWe sampled 3 age‐and‐sex‐matched groups (CU, MCI, and clinical AD) from Mayo Clinic aging studies, each with 61 participants with same‐visit 3D T1‐weighted and 3D T2‐FLAIR MRI, PiB amyloid PET, and Flortaucipir tau PET (total n = 183). We automatically measured hippocampal volume, PIB global SUVR, and tau temporal meta‐ROI SUVR using a) our in‐house pipeline with SPM12, MCALT, and ANTs, and b) FreeSurfer 7.3.2 (PETSurfer), and we measured voxels associated with clinical group separation using SPM12. We also measured WMH from FLAIR using a) our in‐house pipeline, b) Lesion Segmentation Tool (LST), and c) FreeSurfer’s Samseg. For each measurement, we calculated pair‐wise group separation AUC values and Spearman’s rho with age and CDR‐SOB. We then de‐faced all PET and MRI using mri_reface, our top‐performing automated de‐facing software that replaces imaged faces with an average face (“re‐facing”), and we repeated all measurements and analyses with de‐faced images. We then compared the magnitudes of clinical correlations with original images and those of de‐faced images.ResultAnalyses of original and de‐faced images had very high agreement. Only 2/55 paired comparisons were significant (Tables 1‐2): FreeSurfer‐measured global PIB SUVR from de‐faced images was more correlated with age (rho 0.003 vs ‐0.227, p = 0.046), and Mayo‐measured WMH volume from original images was more correlated with CDR‐SOB (rho 0.321 vs 0.308, p = 0.020). Another comparison‐pair tested p = 0.041 for an AUC change +0.008, which we consider not practically meaningful. All others had p>0.05. In voxel‐based analyses of CU vs AD participants, all comparisons were qualitatively similar between original and de‐faced variants (Figure 1).ConclusionDe‐facing PET and MRI using mri_reface did not significantly affect 53/55 correlations between imaging and clinical variables; of the two affected, one became stronger with re‐faced images. Stronger correlations with clinical variables may be explained by de‐facing’s standardizing face voxels and artifacts that irrelevantly affect brain segmentations.

AbstractBackgroundDisruption of default mode network (DMN) connectivity is common in Alzheimer’s disease. We investigated cross‐sectional and one‐year longitudinal changes in DMN sub‐systems in atypical clinical presentations of AD, in relation to age and tau.MethodSixty‐four amyloid‐positive atypical AD (Aty‐AD) patients diagnosed with posterior cortical atrophy (n = 35) or logopenic progressive aphasia (n = 29) were recruited by the Neurodegenerative Research Group. Patients underwent structural and resting‐state functional MRI, and [18F]flortaucipir PET at baseline and one‐year follow‐up. 140 amyloid‐negative cognitively unimpaired (CU) individuals from the Mayo Clinic Study of Aging were included as controls. Global tau SUVR was calculated in a meta‐ROI including parietal, temporal, and occipital regions. DMN connectivity was calculated in native space for posterior (pDMN), ventral (vDMN), anterior ventral (avDMN) and anterior dorsal (adDMN) DMN sub‐systems using the dual regression method. A global measure of DMN connectivity, the Network Failure Quotient (NFQ), was calculated as (vDMN_to_pDMN+ adDMN_to_pDMN)/(pDMN+vDMN). Linear mixed‐effects models were computed for each DMN connectivity measure, including diagnosis (Aty‐AD, CU), age, time and their interaction, and a per‐person random effect. SPM paired t‐tests were run to compare baseline and follow‐up images in Aty‐AD and CU separately. Linear mixed‐effect models were run in the Aty‐AD cohort only to investigate the effect of global baseline tau SUVR on each DMN connectivity measure.ResultMixed‐effects models showed evidence of lower cross‐sectional pDMN and higher adDMN connectivity in Aty‐AD relative to CU, with no significant time effects (Fig. 1). While connectivity in all DMN sub‐systems declined with age in CU, connectivity was higher in older Aty‐AD patients relative to younger, particularly in vDMN and pDMN. A similar trend was observed for NFQ, which was significantly higher in Aty‐AD, with a negative age effect (Fig. 1). In Aty‐AD, voxel‐based analyses showed moderate evidence of decline in vDMN connectivity and increase in avDMN connectivity over time (Fig. 2). Greater baseline global tau SUVR was associated with lower vDMN connectivity, cross‐sectionally and longitudinally (Fig. 3).ConclusionDMN connectivity was altered in Aty‐AD, with strongest alterations in younger patients. The vDMN declined over time, suggesting a shift in DMN connectivity with disease progression that was likely related to tau pathology.