文献速递肿瘤 PET/CT放射性药物与辐射防护

AI变压器模型在超快速肺部PET扫描中的降噪数据高效泛化

Data-efficient generalization of AI transformers for noise reduction in ultra-fast lung PET scans.
PubMed 文献 · Eur J Nucl Med Mol Imaging · 2025-02-26

本研究基于五台扫描仪的1272例回顾性全时PET数据和46例前瞻性超快速20秒屏气(U2BH)PET数据,提出数据高效的掩码视觉变压器(Mask-ViT)方法,在目标扫描仪少量数据微调后直接用于新扫描仪测试。Mask-ViT降噪后的U2BH PET图像在图像质量指标上显著优于U-Net和C-Gan(p<0.05),在三台扫描仪上的病灶检测准确率分别为91.3%、90.4%和91.7%,降噪图像满足病灶可检测性的临床诊断要求。

要点

  • Mask-ViT在目标扫描仪少量数据微调后可直接泛化至新扫描仪,实现数据高效降噪。
  • 降噪后U2BH PET图像质量指标显著优于U-Net和C-Gan(p<0.05)。
  • 三台扫描仪上病灶检测准确率分别为91.3%、90.4%和91.7%。

核心结论

  • Mask-ViT能在数据高效泛化设置下有效提升U2BH PET图像质量,降噪图像满足病灶可检测性的临床诊断要求。

研究设计与样本

  • 研究基于五台扫描仪的两个数据集:第一个数据集包含1272例回顾性收集的全时PET数据,第二个数据集包含46例前瞻性收集的U2BH及对应全时PET/CT图像。
  • 提出数据高效的掩码视觉变压器(Mask-ViT),在目标扫描仪有限训练数据上微调后直接应用于新扫描仪的未见测试数据。
  • 以全时PET图像为参考,将Mask-ViT与U-Net和C-Gan等先进深度学习方法比较,图像质量指标用Wilcoxon符号秩检验进行统计分析。
  • 临床评估由两名读者采用5分制(5=优秀)对图像质量评分,并对诊断质量进行二分类评估。

主要发现

  • Mask-ViT降噪的U2BH PET图像在图像质量指标上较U-Net和C-Gan有统计学显著改善(p<0.05)。
  • 临床评估中,Mask-ViT在三台不同扫描仪上的病灶检测准确率分别为91.3%、90.4%和91.7%。

关键数值与推荐

  • 图像质量评分采用5分制,5分为优秀。
  • 病灶检测准确率在三台扫描仪上分别为91.3%、90.4%和91.7%。

临床怎么用

  • 原文未据此给出推荐。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

PURPOSE: Respiratory motion during PET acquisition may produce lesion blurring. Ultra-fast 20-second breath-hold (U2BH) PET reduces respiratory motion artifacts, but the shortened scanning time increases statistical noise and may affect diagnostic quality. This study aims to denoise the U2BH PET images using a deep learning (DL)-based method. METHODS: The study was conducted on two datasets collected from five scanners where the first dataset included 1272 retrospectively collected full-time PET data while the second dataset contained 46 prospectively collected U2BH and the corresponding full-time PET/CT images. A robust and data-efficient DL method called mask vision transformer (Mask-ViT) was proposed which, after fine-tuned on a limited number of training data from a target scanner, was directly applied to unseen testing data from new scanners. The performance of Mask-ViT was compared with state-of-the-art DL methods including U-Net and C-Gan taking the full-time PET images as the reference. Statistical analysis on image quality metrics were carried out with Wilcoxon signed-rank test. For clinical evaluation, two readers scored image quality on a 5-point scale (5 = excellent) and provided a binary assessment for diagnostic quality evaluation. RESULTS: The U2BH PET images denoised by Mask-ViT showed statistically significant improvement over U-Net and C-Gan on image quality metrics (p < 0.05). For clinical evaluation, Mask-ViT exhibited a lesion detection accuracy of 91.3%, 90.4% and 91.7%, when it was evaluated on three different scanners. CONCLUSION: Mask-ViT can effectively enhance the quality of the U2BH PET images in a data-efficient generalization setup. The denoised images meet clinical diagnostic requirements of lesion detectability.

来源:Eur J Nucl Med Mol Imaging,查看原文。版权归原作者所有。

引用

  1. Wang J, Zhang X, Miao Y, et al. Data-efficient generalization of AI transformers for noise reduction in ultra-fast lung PET scans. Eur J Nucl Med Mol Imaging. 2025. PMID: 40009163. doi:10.1007/s00259-025-07165-7

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。