PET-Adapter: Test-Time Domain Adaptation for Full and Limited-Angle PET Image Reconstruction

2026-05-08

AI Summary

This work addresses the limited generalization of deep learning–based PET image reconstruction models under unseen clinical data distributions by proposing PET-Adapter, a test-time domain adaptation framework for cross-domain reconstruction. Integrating a layer-wise low-rank anatomical conditioning mechanism with an OSEM-based physics-guided warm-start strategy, PET-Adapter effectively adapts to diverse anatomies, tracers, and scanner configurations without requiring paired ground-truth data. The method substantially enhances reconstruction quality in both full-angle and limited-angle scenarios, reduces the number of diffusion steps from 50 to 2—significantly lowering computational cost—and maintains strong 3D reconstruction performance and clinical feasibility across multiple real-world clinical datasets.

Abstract

Positron Emission Tomography (PET) image reconstruction is inherently challenged by Poisson noise and physical degradation factors, which are further exacerbated in limited-angle acquisitions. While deep learning methods demonstrate promising performance, their generalization to unseen clinical data distributions remains limited without extensive retraining. We propose PET-Adapter, a test-time domain adaptation framework for generative PET reconstruction models pretrained solely on phantom data. Our method enables adaptation to clinical datasets with varying anatomies, tracers, and scanner configurations without requiring paired ground truth. PET-Adapter introduces layer-wise low-rank anatomical conditioning during adaptation and Ordered Subset Expectation Maximization-based warm-starting that initializes the generation from physics-informed reconstructions, reducing diffusion steps from 50 to 2 without compromising quality. Experiments across multiple clinical datasets demonstrate superior 3D reconstruction performance in both full-angle and limited-angle settings, highlighting the clinical feasibility and computational efficiency of the proposed approach.

Problem

• PET image reconstruction
• domain adaptation
• limited-angle acquisition
• generalization
• clinical data distribution

Innovation

• test-time domain adaptation
• low-rank anatomical conditioning
• diffusion model acceleration
• physics-informed warm-starting
• limited-angle PET reconstruction