用于神经内分泌肿瘤患者[64Cu]Cu-DOTATATE PET/CT全肿瘤分割的卷积神经网络
研究纳入胃肠胰或肺神经内分泌肿瘤患者(训练集117例、测试集41例)及10例无神经内分泌肿瘤征象的阴性对照,采用nnU-Net框架构建深度学习U-net架构,以医师标准化半自动分割为参考标准。集成模型病灶层面Dice为0.850,像素层面Dice、精确率、灵敏度分别为0.801、0.786、0.872;35/41(85%)患者经一定程度手动调整后AI分割可接受,最终分割耗时5分钟 vs 参考方法17分钟,p<0.01。
要点
- 集成模型病灶层面Dice为0.850,像素层面Dice、精确率、灵敏度分别为0.801、0.786、0.872。
- 35/41(85%)患者经一定程度手动调整后AI分割可接受。
- AI模型加手动调整获得最终肿瘤分割耗时5分钟 vs 参考方法17分钟,p<0.01。
核心结论
- 研究实现并验证了一个与参考分割高度相似的AI模型,使肿瘤分割更快。
- 作者认为借助AI,全肿瘤分割可能变得适用于临床常规。
研究设计与样本
- 纳入胃肠胰或肺神经内分泌肿瘤患者,训练集117例、测试集41例,另纳入10例无神经内分泌肿瘤征象的阴性对照。
- 以医师标准化半自动方法获得的肿瘤分割作为参考标准,采用nnU-Net框架构建深度学习U-net架构。
- 以Dice评分、灵敏度和精确率选择最终模型,并在临床影像查看器中由医师评估AI分割并手动调整。
主要发现
- 集成模型总体表现最佳,病灶层面Dice为0.850,像素层面Dice、精确率、灵敏度分别为0.801、0.786、0.872。
- 集成模型在35/41(85%)患者中经一定程度手动调整后表现可接受。
- AI模型加手动调整获得最终肿瘤分割耗时5分钟,参考方法为17分钟,p<0.01。
关键数值与推荐
- 模型选择指标为Dice评分、灵敏度和精确率。
- 集成模型病灶层面Dice 0.850,像素层面Dice 0.801、精确率0.786、灵敏度0.872。
- AI分割可接受者35/41(85%),最终分割耗时5分钟 vs 17分钟,p<0.01。
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
BACKGROUND: Segmentation of neuroendocrine neoplasms (NENs) in [64Cu]Cu-DOTATATE positron emission tomography makes it possible to extract quantitative measures useable for prognostication of patients. However, manual tumor segmentation is cumbersome and time-consuming. Therefore, we aimed to implement and test an artificial intelligence (AI) network for tumor segmentation. Patients with gastroenteropancreatic or lung NEN with [64Cu]Cu-DOTATATE PET/CT performed were included in our training (n = 117) and test cohort (n = 41). Further, 10 patients with no signs of NEN were included as negative controls. Ground truth segmentations were obtained by a standardized semiautomatic method for tumor segmentation by a physician. The nnU-Net framework was used to set up a deep learning U-net architecture. Dice score, sensitivity and precision were used for selection of the final model. AI segmentations were implemented in a clinical imaging viewer where a physician evaluated performance and performed manual adjustments. RESULTS: Cross-validation training was used to generate models and an ensemble model. The ensemble model performed best overall with a lesion-wise dice of 0.850 and pixel-wise dice, precision and sensitivity of 0.801, 0.786 and 0.872, respectively. Performance of the ensemble model was acceptable with some degree of manual adjustment in 35/41 (85%) patients. Final tumor segmentation could be obtained from the AI model with manual adjustments in 5 min versus 17 min for ground truth method, p < 0.01. CONCLUSION: We implemented and validated an AI model that achieved a high similarity with ground truth segmentation and resulted in faster tumor segmentation. With AI, total tumor segmentation may become feasible in the clinical routine.
引用
- Carlsen EA, Lindholm K, Hindsholm A, et al. A convolutional neural network for total tumor segmentation in [64Cu]Cu-DOTATATE PET/CT of patients with neuroendocrine neoplasms. EJNMMI Res. 2022. PMID: 35633448. doi:10.1186/s13550-022-00901-2
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。