Fast Voxel-Wise Kinetic Modeling in Dynamic PET using a Physics-Informed CycleGAN

2025-10-27

AI Summary

Dynamic PET voxel-wise kinetic modeling has long been hindered by the invasiveness and high inter-subject variability of arterial input function (AIF) estimation. To address this, we propose a physics-guided CycleGAN framework that, for the first time, embeds biophysical priors into unpaired image-to-image translation, enabling end-to-end, noninvasive prediction of both AIF and parametric maps (e.g., K₁, Vₜ) directly from dynamic PET sequences. Our method jointly enforces PET forward-physics modeling and cycle-consistency constraints, eliminating the need for invasive blood sampling or explicit compartmental modeling. Validated on multicenter data, our approach achieves strong agreement between predicted and gold-standard AIFs (r > 0.92) and high correlation with conventional nonlinear fitting results for kinetic parameters (r > 0.95 for K₁ and Vₜ). Moreover, computation time is reduced by two orders of magnitude. This work establishes a novel, rapid, and generalizable paradigm for quantitative dynamic PET analysis.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Cognitive Modeling & Cognitive Systems: Affective Computing
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Tracer kinetic modeling serves a vital role in diagnosis, treatment planning, tracer development and oncology, but burdens practitioners with complex and invasive arterial input function estimation (AIF). We adopt a physics-informed CycleGAN showing promise in DCE-MRI quantification to dynamic PET quantification. Our experiments demonstrate sound AIF predictions and parameter maps closely resembling the reference.

Problem

• Estimating arterial input function without invasive procedures
• Improving kinetic modeling accuracy in dynamic PET imaging
• Applying physics-informed CycleGAN for medical image quantification

Innovation

• Physics-informed CycleGAN for dynamic PET quantification
• Non-invasive arterial input function estimation method
• Generates accurate kinetic parameter maps without AIF