Artificial Intelligence–Powered Quantification of Flortaucipir PET for Detecting Tau Pathology
Bibliographic record
Abstract
We developed and evaluated an artificial intelligence (AI)–powered approach for easier quantification of tau PET uptake without requiring structural MR to aid earlier tracking of Alzheimer disease (AD). Methods: We implemented a deep neural network model that normalizes 18F-AV1451 (tau) PET images to a standard template without requiring MR, using transfer learning from a model pretrained on amyloid PET. This model was integrated into an MR-free pipeline for tau PET quantification and validated on external dataset (Alzheimer Disease Neuroimaging Initiative). We examined correlations between model-derived tau uptake estimates and cognitive measures, including AD stage and episodic memory performance (n = 666). Longitudinal analyses were conducted to assess whether baseline tau deposition predicted future cognitive decline (n = 168). Results: The AI-powered pipeline achieved robust performance with intraclass correlation coefficients exceeding 0.97 for regional uptake estimation compared with MR-based ground truth. We also showed that the tau deposition in metatemporal regions was significantly correlated with Mini-Mental State Examination and Montreal Cognitive Assessment scores. Elevated tau PET uptake in the entorhinal cortex and inferior temporal gyrus predicted future cognitive decline. Conclusion: The proposed AI-powered pipeline enhances the clinical accessibility of tau PET by reducing scan costs and streamlining the uptake quantification, achieving high performance without requiring structural MR. We further demonstrated that the pipeline yields cognitively relevant outcome measures for early diagnosis and monitoring of AD progression to aid more personalized treatment strategies targeting AD biomarkers.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".