Feynman Kac Reweighted Schrödinger Bridge Matching for Surface-Based Tau PET Harmonization

2026-06-16 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Medical Imaging AI · Depth: Expert, quick

Summary

The Feynman Kac Reweighted Schrödinger Bridge Matching (FKRSBM) model addresses nonbiological variability in Tau PET imaging data, a critical issue for tracking Alzheimer's disease progression. Existing harmonization methods often conflate site effects with biological variation, especially when source and target cohorts differ in subgroup composition. FKRSBM learns a direct stochastic transport process between distributions using entropy-regularized optimal transport, avoiding Gaussian noise priors. It incorporates a subgroup-aware endpoint proposal via Feynman Kac reweighting and stratified importance sampling, without altering the underlying solver or network. For surface-based neuroimaging, FKRSBM employs a spherical convolutional backbone. Evaluated on PI-2620 data (HABS-HD) harmonized into the AV-1451 domain (ADNI), FKRSBM achieved superior distributional alignment, reduced tau-positivity sign mismatch, stronger APOE subgroup alignment, and improved downstream disease classification compared to ComBat, CycleGAN, a diffusion-based method, and unregularized Diffusion Schrödinger Bridge Matching.

Key takeaway

For research scientists analyzing Tau PET data, existing harmonization methods risk conflating site effects with crucial biological variation, potentially biasing Alzheimer's disease assessments. You should consider the FKRSBM model to achieve superior distributional alignment and preserve biological signals, especially when working with diverse cohorts. Implementing FKRSBM can lead to more accurate tau-positivity detection and improved downstream disease classification performance.

Key insights

FKRSBM directly harmonizes Tau PET data, preserving biological signals despite subgroup differences.

Principles

• Direct stochastic transport improves harmonization.
• Subgroup-aware reweighting prevents biological signal loss.
• Entropy-regularized optimal transport is effective.

Method

FKRSBM learns direct stochastic transport via entropy-regularized optimal transport, using a subgroup-aware endpoint proposal from Feynman Kac reweighting, implemented with stratified importance sampling and a spherical convolutional backbone.

In practice

• Harmonize PI-2620 data to AV-1451 domain.
• Apply spherical convolutions for surface-based neuroimaging.
• Improve Alzheimer's disease classification.

Topics

• Tau PET Imaging
• Data Harmonization
• Schrödinger Bridge Matching
• Optimal Transport
• Alzheimer's Disease
• Neuroimaging

Best for: AI Scientist, Research Scientist

Related on AIssential

• Transition-Based Digital Twin Modelling for Alzheimer's Disease under Sparse Longitudinal Data — Local transition modeling offers a data-efficient and robust strategy for personalized disease forecasting in neurodegenerative disorders.
• Geometric and Quantum Kernel Methods for Predicting Skeletal Muscle Outcomes in chronic obstructive pulmonary disease — Quantum and geometric kernel methods improve low-data biomedical prediction by capturing non-linear feature interactions.
• A physics-informed graph neural network to approximate docking-based binding affinity for DYRK2 in Alzheimer’s drug repurposing — Integrating physical interaction terms into GNNs can create efficient, interpretable surrogates for computationally intensive molecular docking.
• Improved Multiscale Structural Mapping with Supervertex Vision Transformer for the Detection of Alzheimer's Disease Neurodegeneration — MSSM+ and SV-ViT enhance AD detection via MRI by integrating diverse cortical features and supervertex-based learning.
• Reusability report: Meta-learning for antigen-specific T cell receptor binder identification — PanPep generalizes well to unseen antigens with few known binders under specific negative sampling.
• Bayesian In Vivo Tracking of Synapses using Joint Poisson Deconvolution and Diffeomorphic Registration — A Bayesian framework enables robust, simultaneous synapse tracking, denoising, and registration in noisy in vivo microscopy data.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.