基于深度学习的前列腺癌原发灶PSMA-PET前列腺内大体肿瘤体积自动勾画
本研究训练3D U-Net在128例来自三个机构的18F-PSMA-1007 PET图像上自动勾画前列腺内大体肿瘤体积,并在52例内部和外部独立队列中测试,与专家勾画相比中位Dice系数为0.71-0.82,敏感性和特异性与专家接近,单例预测平均耗时3.81秒。
要点
- 3D U-Net在128例18F-PSMA-1007 PET上训练,在52例内部和外部队列测试,中位Dice系数0.71-0.82。
- CNN与专家勾画的敏感性中位0.88 vs 0.85(p=0.40),特异性中位0.83 vs 0.88(p=0.014)。
- CNN预测平均每例耗时3.81秒,GTV体积与专家勾画无显著差异(所有比较p>0.1)。
核心结论
- CNN在内部和外部数据集及组织病理学参考上训练和测试,实现了三种PSMA-PET示踪剂的快速GTV分割,诊断准确性与手动专家相当。
研究设计与样本
- 研究为深度学习模型开发与验证,训练集为来自三个机构的128例18F-PSMA-1007 PET图像,测试集为52例患者,包括一个内部独立队列(Freiburg:n=19)和三个外部独立队列(Dresden:n=14例18F-PSMA-1007;Boston:Massachusetts General Hospital(MGH):n=9例18F-DCFPyL-PSMA;Dana-Farber Cancer Institute(DFCI):n=10例68Ga-PSMA-11)。
- 专家轮廓由共识使用验证技术生成,CNN预测与专家轮廓使用Dice相似系数(DSC)比较,内部测试队列使用配准的全mount组织学评估敏感性/特异性。
主要发现
- 中位DSC分别为Freiburg:0.82(IQR:0.73-0.88),Dresden:0.71(IQR:0.53-0.75),MGH:0.80(IQR:0.64-0.83),DFCI:0.80(IQR:0.67-0.84)。
- CNN和专家轮廓的中位敏感性分别为0.88(IQR:0.68-0.97)和0.85(IQR:0.75-0.88)(p=0.40)。
- GTV体积无显著差异(所有比较p>0.1)。
- CNN和专家轮廓的中位特异性分别为0.83(IQR:0.57-0.97)和0.88(IQR:0.69-0.98)(p=0.014)。
- CNN预测平均每例耗时3.81秒。
关键数值与推荐
- 中位DSC分别为Freiburg:0.82(IQR:0.73-0.88),Dresden:0.71(IQR:0.53-0.75),MGH:0.80(IQR:0.64-0.83),DFCI:0.80(IQR:0.67-0.84)。
- CNN和专家轮廓的中位敏感性分别为0.88(IQR:0.68-0.97)和0.85(IQR:0.75-0.88)(p=0.40)。
- CNN和专家轮廓的中位特异性分别为0.83(IQR:0.57-0.97)和0.88(IQR:0.69-0.98)(p=0.014)。
- GTV体积无显著差异(所有比较p>0.1)。
- CNN预测平均每例耗时3.81秒。
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: With the increased use of focal radiation dose escalation for primary prostate cancer (PCa), accurate delineation of gross tumor volume (GTV) in prostate-specific membrane antigen PET (PSMA-PET) becomes crucial. Manual approaches are time-consuming and observer dependent. The purpose of this study was to create a deep learning model for the accurate delineation of the intraprostatic GTV in PSMA-PET. METHODS: A 3D U-Net was trained on 128 different 18F-PSMA-1007 PET images from three different institutions. Testing was done on 52 patients including one independent internal cohort (Freiburg: n = 19) and three independent external cohorts (Dresden: n = 14 18F-PSMA-1007, Boston: Massachusetts General Hospital (MGH): n = 9 18F-DCFPyL-PSMA and Dana-Farber Cancer Institute (DFCI): n = 10 68Ga-PSMA-11). Expert contours were generated in consensus using a validated technique. CNN predictions were compared to expert contours using Dice similarity coefficient (DSC). Co-registered whole-mount histology was used for the internal testing cohort to assess sensitivity/specificity. RESULTS: Median DSCs were Freiburg: 0.82 (IQR: 0.73-0.88), Dresden: 0.71 (IQR: 0.53-0.75), MGH: 0.80 (IQR: 0.64-0.83) and DFCI: 0.80 (IQR: 0.67-0.84), respectively. Median sensitivity for CNN and expert contours were 0.88 (IQR: 0.68-0.97) and 0.85 (IQR: 0.75-0.88) (p = 0.40), respectively. GTV volumes did not differ significantly (p > 0.1 for all comparisons). Median specificity of 0.83 (IQR: 0.57-0.97) and 0.88 (IQR: 0.69-0.98) were observed for CNN and expert contours (p = 0.014), respectively. CNN prediction took 3.81 seconds on average per patient. CONCLUSION: The CNN was trained and tested on internal and external datasets as well as histopathology reference, achieving a fast GTV segmentation for three PSMA-PET tracers with high diagnostic accuracy comparable to manual experts.
引用
- Holzschuh JC, Mix M, Ruf J, et al. Deep learning based automated delineation of the intraprostatic gross tumour volume in PSMA-PET for patients with primary prostate cancer. Radiother Oncol. 2023. PMID: 37394103. doi:10.1016/j.radonc.2023.109774
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。