Fed-NDIF: A Noise-Embedded Federated Diffusion Model For Low-Count Whole-Body PET Denoising

2025-03-20

AI Summary

Low-count PET (LCPET) imaging reduces radiation exposure but suffers from high noise and poor lesion detectability. Existing diffusion models require large-scale annotated datasets, limiting their applicability in multi-center medical settings due to data silos, scanner heterogeneity, and strict privacy constraints. To address this, we propose Fed-NDIF—a noise-embedded federated diffusion model—introducing liver-normalized standard deviation (NSTD) as a noise prior within a 2.5D diffusion framework. Fed-NDIF integrates FedAvg with personalized local fine-tuning to enable collaborative training across heterogeneous, privacy-sensitive multi-center data. Evaluated on LCPET datasets from the University of Bern, Ruijin Hospital, and Yale New Haven Hospital, Fed-NDIF consistently outperforms local diffusion models and federated U-Net, achieving significant improvements in PSNR, SSIM, NMSE, lesion detection rate, and SUV quantification accuracy. This work establishes a novel privacy-preserving paradigm for low-dose PET denoising.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Distributed Machine Learning & Federated Learning
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Privacy-enhancing technologies
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing

Abstract

Low-count positron emission tomography (LCPET) imaging can reduce patients' exposure to radiation but often suffers from increased image noise and reduced lesion detectability, necessitating effective denoising techniques. Diffusion models have shown promise in LCPET denoising for recovering degraded image quality. However, training such models requires large and diverse datasets, which are challenging to obtain in the medical domain. To address data scarcity and privacy concerns, we combine diffusion models with federated learning -- a decentralized training approach where models are trained individually at different sites, and their parameters are aggregated on a central server over multiple iterations. The variation in scanner types and image noise levels within and across institutions poses additional challenges for federated learning in LCPET denoising. In this study, we propose a novel noise-embedded federated learning diffusion model (Fed-NDIF) to address these challenges, leveraging a multicenter dataset and varying count levels. Our approach incorporates liver normalized standard deviation (NSTD) noise embedding into a 2.5D diffusion model and utilizes the Federated Averaging (FedAvg) algorithm to aggregate locally trained models into a global model, which is subsequently fine-tuned on local datasets to optimize performance and obtain personalized models. Extensive validation on datasets from the University of Bern, Ruijin Hospital in Shanghai, and Yale-New Haven Hospital demonstrates the superior performance of our method in enhancing image quality and improving lesion quantification. The Fed-NDIF model shows significant improvements in PSNR, SSIM, and NMSE of the entire 3D volume, as well as enhanced lesion detectability and quantification, compared to local diffusion models and federated UNet-based models.

Problem

• Denoising low-count PET images to reduce radiation exposure
• Addressing data scarcity and privacy in medical imaging with federated learning
• Improving lesion detectability and quantification in multicenter PET datasets

Innovation

• Combines diffusion models with federated learning
• Uses noise-embedded 2.5D diffusion model
• Applies Federated Averaging for model aggregation