Synthetic CT Generation from Time-of-Flight Non-Attenutaion-Corrected PET for Whole-Body PET Attenuation Correction

2025-04-10

AI Summary

To address inaccurate attenuation correction (AC) in PET/MR systems due to the absence of CT, this work proposes a novel method for synthesizing CT (sCT) directly from time-of-flight (TOF) non-attenuation-corrected (NAC) PET images. The approach innovatively adapts natural-image pre-trained models to medical image translation via a “natural-image pretraining + few-shot medical fine-tuning” strategy, jointly optimizing TOF-PET feature extraction and CT Hounsfield unit (HU) space modeling. Evaluated on only 35 clinical cases, the method achieves high-fidelity, concurrent reconstruction of bone and soft-tissue structures, yielding an intra-body MAE of 74.49 HU and PSNR of 28.66 dB—substantially outperforming models trained exclusively on medical data. Qualitative assessment confirms marked improvements in bony detail delineation and soft-tissue contrast.

Technology Category

• Computer Vision: Medical and Biological Imaging
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Positron Emission Tomography (PET) imaging requires accurate attenuation correction (AC) to account for photon loss due to tissue density variations. In PET/MR systems, computed tomography (CT), which offers a straightforward estimation of AC is not available. This study presents a deep learning approach to generate synthetic CT (sCT) images directly from Time-of-Flight (TOF) non-attenuation corrected (NAC) PET images, enhancing AC for PET/MR. We first evaluated models pre-trained on large-scale natural image datasets for a CT-to-CT reconstruction task, finding that the pre-trained model outperformed those trained solely on medical datasets. The pre-trained model was then fine-tuned using an institutional dataset of 35 TOF NAC PET and CT volume pairs, achieving the lowest mean absolute error (MAE) of 74.49 HU and highest peak signal-to-noise ratio (PSNR) of 28.66 dB within the body contour region. Visual assessments demonstrated improved reconstruction of both bone and soft tissue structures from TOF NAC PET images. This work highlights the effectiveness of using pre-trained deep learning models for medical image translation tasks. Future work will assess the impact of sCT on PET attenuation correction and explore additional neural network architectures and datasets to further enhance performance and practical applications in PET imaging.

Problem

• Generate synthetic CT from TOF NAC PET for PET/MR attenuation correction
• Improve PET imaging accuracy by deep learning-based attenuation correction
• Enhance bone and soft tissue reconstruction using pre-trained models

Innovation

• Deep learning generates synthetic CT from PET
• Pre-trained models enhance medical image translation
• Fine-tuning achieves high accuracy in CT reconstruction