文献速递肿瘤 PET/CT

基于数据中心深度学习的[18F]FDG与[68Ga]/[18F]PSMA PET/CT全自动分割

Fully Automated Segmentation of [18F]FDG and [68Ga]/[18F]PSMA PET/CT Images via Data-Centric Deep Learning.
PubMed 文献 · J Nucl Med · 2026-09-30

本研究开发并验证了开源PET-only肿瘤分割流程LION,用于[18F]FDG和PSMA靶向PET/CT,并探讨训练数据特征对分割性能的影响。回顾性多中心研究使用5209例[18F]FDG和2046例PSMA靶向PET/CT训练模型,在多中心留出队列评估。器官背景类别使[18F]FDG中位Dice从0.62提升至0.71,PSMA从0.75提升至0.83;肿瘤空间多样性(TOM)与Dice强相关(ρ=0.80,P=0.003)。LION在[18F]FDG和PSMA中位Dice分别为0.71和0.85,优于其他开源工具。

要点

  • LION实现仅需PET的[18F]FDG与PSMA靶向PET/CT全自动肿瘤分割。
  • 引入器官生理摄取作为辅助类别可提升分割性能。
  • 训练数据空间多样性(TOM量化)与分割Dice强相关。

核心结论

  • LION可实现[18F]FDG和PSMA靶向PET的PET-only自动分割,训练数据组成尤其是TOM量化的空间多样性与分割性能强相关。

研究设计与样本

  • 回顾性多中心研究,使用5209例[18F]FDG PET/CT(涵盖19种疾病类型)和2046例PSMA靶向PET/CT训练PET-only分割模型。
  • 在多中心留出队列评估:616例[18F]FDG扫描(涵盖4种疾病)和443例PSMA靶向前列腺癌扫描,并与3种开源工具比较。

主要发现

  • 在完整留出队列中,器官背景使[18F]FDG中位Dice从0.62提升至0.71,PSMA从0.75提升至0.83。
  • TOM测量的空间多样性与Dice强相关(Spearman ρ=0.80,P=0.003)。
  • 在500例患者上训练的混合疾病模型性能与在3031例上训练的淋巴瘤专科模型相当。
  • DINOv2嵌入揭示同疾病队列间存在扫描仪诱导的领域偏移。
  • LION在[18F]FDG和PSMA的中位Dice分别为0.71和0.85,在排除AutoPET测试病例的模型比较测试集上优于其他开源方法。

临床怎么用

  • LION可作为开源PET-only分割工具用于[18F]FDG和PSMA靶向PET/CT的肿瘤自动识别。
  • 训练分割模型时应关注数据空间多样性(可用TOM量化),而非仅增加病例数。

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

The purpose of this study was to develop and validate lesion identification in oncologic nuclear imaging (LION), an open-source PET-only tumor segmentation pipeline for [18F]FDG and prostate-specific membrane antigen (PSMA)-targeted PET/CT, and to investigate how training data characteristics influence segmentation performance. Methods: In this retrospective multicenter study, 5209 [18F]FDG PET/CT scans spanning 19 disease types and 2046 PSMA-targeted PET/CT scans were used to train PET-only segmentation models. Tumor segmentation incorporated organs with physiologic uptake as auxiliary classes to enable PET-only inference. Tumor occurrence maps (TOMs) quantified tumor spatial diversity across the training data. For [18F]FDG, disease-specific and mixed-disease models trained on progressively larger subsets were compared to test whether increasing spatial diversity improves generalization. Scanner-related domain shift was analyzed using DINOv2 embeddings. Models were evaluated on multicenter holdout cohorts (616 [18F]FDG scans across 4 diseases; 443 PSMA-targeted prostate cancer scans) and compared with 3 open-source tools. Results: Organ context improved median Dice from 0.62 to 0.71 for [18F]FDG and from 0.75 to 0.83 for PSMA, on the complete holdout cohorts. Spatial diversity measured by TOMs was strongly associated with Dice (Spearman ρ = 0.80, P = 0.003). A mixed-disease model trained on 500 patients matched the performance of a lymphoma specialist model trained on 3031 cases. DINOv2 embeddings revealed scanner-induced domain shift between same-disease cohorts. LION achieved median Dice scores of 0.71 for [18F]FDG and 0.85 for PSMA and outperformed other open-source approaches on the model-comparison test set, which excluded the AutoPET test cases. Conclusion: LION enables PET-only automated segmentation for [18F]FDG and PSMA-targeted PET. Training data composition, particularly spatial diversity quantified by TOMs, was strongly associated with segmentation performance.

来源:J Nucl Med,查看原文。版权归原作者所有。

引用

  1. Pires M, Gutschmayer S, Bergalla E, et al. Fully Automated Segmentation of [18F]FDG and [68Ga]/[18F]PSMA PET/CT Images via Data-Centric Deep Learning. J Nucl Med. 2026. PMID: 42816125. doi:10.2967/jnumed.126.272852

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。