Fig. 4: Association between pointwise medial temporal lobe thickness and ipsilateral, quantitative NFT burden measures.

The t-statistic maps show the correlation between pointwise cortical thickness and the mean NFT burden computed within different anatomical subregions (n = 25). The NFT burden measures are computed based on the 3-D quantitative NFT density maps. The analyses are arranged in the order in which the anatomical subregions used are affected during the AD process, from early to late, based on the results shown in Fig. 2C. The t-statistic maps reveal increasingly strengthened associations with NFT burden measures derived from regions affected during the later stage of AD. Clusters were defined based on an empirical threshold (uncorrected p < 0.01) and permutation testing with the Freedman & Lane method (1000 iterations) was used to assign each potential cluster a one-sided, corrected p-value. To account for multiple comparisons, the analysis uses cluster-level family-wise error rate correction. The clusters outlined in black indicate regions where a significant correlation was observed after correction for multiple hypothesis testing (corrected p < 0.05). No covariates were included in this model. (S subiculum, PrS-PaS Pre/Parasubiculum, SRLM = stratum radiatum lacunosum molecular, PP perforant pathway, CA cornu ammonis, DG dentate gyrus, ERC entorhinal cortex, BA Brodmann area).