Fig. 5
From: A deep learning MRI approach outperforms other biomarkers of prodromal Alzheimer’s disease
The deep learning MRI score correlates with tau pathology. The scatter plots illustrate the relationship between changes over time in the deep learning MRI scores vs. changes in CSF Aβ (left panel), changes in CSF tau (middle panel), and changes in CSF tau/Aβ (right panel). Each data point indicates one participant’s change of last deep learning MRI score from baseline (ΔDLMRIlast), plotted against their fitted change in biomarker measures at ΔDLMRIlast with the slope estimated from all follow-up visits (see the “Methods” section). The black solid lines are the linear fits across participants, showing that changes in the deep learning MRI score are most strongly correlated with the changes in tau over time. The table lists the correlations between antemortem deep learning MRI scores to postmortem-derived Braak stage of neurofibrillary tangles and the Thal phase of amyloid plaques, with an MRI autopsy interval below either 1 or 2 years, showing that the deep learning MRI scores are most strongly correlated with tau pathology