Brain tau PET-based identification and characterization of subpopulations in patients with Alzheimer's disease using deep learning-derived saliency maps

Abstract

Background: Alzheimer's disease (AD) is a heterogeneous neurodegenerative disorder in which tau neurofibrillary tangles are a pathological hallmark closely associated with cognitive dysfunction and neurodegeneration. In this study, we used brain tau data to investigate AD heterogeneity by identifying and characterizing the subpopulations among patients. We included 615 cognitively normal and 159 AD brain ¹⁸F-flortaucipr PET scans, along with T1-weighted MRI from the Alzheimer Disease Neuroimaging Initiative database. A three dimensional-convolutional neural network model was employed for AD detection using standardized uptake value ratio (SUVR) images. The model-derived saliency maps were generated and employed as informative image features for clustering AD participants. Among the identified subpopulations, statistical analysis of demographics, neuropsychological measures, and SUVR were compared. Correlations between neuropsychological measures and regional SUVRs were assessed. A generalized linear model was utilized to investigate the sex and APOE ε4 interaction effect on regional SUVRs.

Results: Two distinct subpopulations of AD patients were revealed, denoted as S_Hi and S_Lo. Compared to the S_Lo group, the S_Hi group exhibited a significantly higher global tau burden in the brain, but both groups showed similar cognition distribution levels. In the S_Hi group, the associations between the neuropsychological measurements and regional tau deposition were weakened. Moreover, a significant interaction effect of sex and APOE ε4 on tau deposition was observed in the S_Lo group, but no such effect was found in the S_Hi group.

Conclusion: Our results suggest that tau tangles, as shown by SUVR, continue to accumulate even when cognitive function plateaus in AD patients, highlighting the advantages of PET in later disease stages. The differing relationships between cognition and tau deposition, and between gender, APOE4, and tau deposition, provide potential for subtype-specific treatments. Targeting gender-specific and genetic factors influencing tau deposition, as well as interventions aimed at tau's impact on cognition, may be effective.

Keywords: Alzheimer’s disease; Apolipoprotein E; Cognitive; Deep learning; Heterogeneity; Neuropsychological measures; Saliency map; Tau PET.

Fig. 1

Schematic illustration of workflow for identification and characterization of subpopulations in AD. (A) The imaging data including ¹⁸F-flortaucipir PET and T1-weighted MRI and clinical data were collected from the ADNI cohort, and processed for being used in the following steps of (B) to (D). (B) The 3D-CNN model construction and evaluation for AD and CN classification. (C) The developed 3D-CNN model was used to identify subpopulations in AD participants. (D) Two identified subpopulations from (C) were then characterized by investigating the relationships between the clinical factors and ¹⁸F-flortaucipir tau depositions

Fig. 2

ROC analysis for the 3D-CNN model. The 3D-CNN model outperformed two conventional meta-ROI ¹⁸F-flortaucipir SUVR approaches for CN and AD predication in terms of the area under the ROC curve

Fig. 3

Mean of saliency maps and tau SUVR images, and statistical t-maps for group differences. (A) The mean saliency maps for participants in AD subpopulation 1 (S_Lo), AD subpopulation 2 (S_Hi) and the statistical t map from the voxel-wise comparison. This parametric map was set at threshold of p < 0·001. (B) The mean tau SUVR images for participants in the S_Lo, S_Hi, and the statistical t map from the voxel-wise comparison. This parametric map was set at a threshold at p < 0·001

Fig. 4

Correlation analyses between ¹⁸F-flortaucipir ROI SUVRs and neuropsychological measures (FAQ, ADNI-MEM, ADNI-EF) in the S_Lo and S_Hi groups. The regions analyzed are: (A), (D), and (G) amygdala; (B), (E), and (H) medial temporal; (C), (F), and (I) entorhinal cortex

Fig. 5

The APOE ε4 and sex interaction effect on ROI SUVRs in S_Lo and S_Hi groups. Representative ROIs are the amygdala (A), medial temporal (B), entorhinal cortex (C), inferior temporal (D), parahippocampal (E), and lateral temporal (F) regions

Fig. 6

Comparison of the feature importance across brain regions on PET SUVR and saliency map