PDF(Pubmed)
Abstract：
BACKGROUND: Models predicting future macrovascular invasion in hepatocellular carcinoma are constructed to assist timely interventions.
METHODS: A total of 366 HCC cases were retrospectively collected from five Chinese hospitals between April 2007 and November 2016: the training dataset comprised 281 patients from four hospitals; the external validation dataset comprised 85 patients from another hospital. Multi-task deep learning network-based models were constructed to predict future macrovascular invasion. The discrimination, calibration, and decision curves were compared to identify the best model. We compared the time to macrovascular invasion and overall survival using the best model and related image heterogeneity scores (H-score). Then, we determined the need for a segmentation subnet or the replacement deep learning algorithm by logistic regression in screening clinical/radiological factors. Finally, an applet was constructed for future application.
RESULTS: The best model combined clinical/radiological factors and radiomic features. It achieved best discrimination (areas under the curve: 0·877 in the training dataset and 0·836 in the validation dataset), calibration, and decision curve. Its performance was not affected by the treatments and disease stages. The subgroups had statistical significance for time to macrovascular invasion (training: hazard ratio [HR] = 0·073, 95% confidence interval [CI]: 0·032-0·167, p < 0·001 and validation: HR = 0·090, 95%CI: 0·022-0·366, p < 0·001) and overall survival (training: HR = 0·344, 95%CI: 0·246-0·547, p < 0·001 and validation: HR = 0·489, 95%CI: 0·279 - 0·859, p = 0·003). Similar results were achieved when the patients were subdivided by the H-score. The subnet for segmentation and end-to-end deep learning algorithms improved the performance of the model.
CONCLUSIONS: Our multi-task deep learning network-based model successfully predicted future macrovascular invasion. In high-risk populations, besides the current first-line treatments, more therapies may be explored for macrovascular invasion.
METHODS: A total of 366 HCC cases were retrospectively collected from five Chinese hospitals between April 2007 and November 2016: the training dataset comprised 281 patients from four hospitals; the external validation dataset comprised 85 patients from another hospital. Multi-task deep learning network-based models were constructed to predict future macrovascular invasion. The discrimination, calibration, and decision curves were compared to identify the best model. We compared the time to macrovascular invasion and overall survival using the best model and related image heterogeneity scores (H-score). Then, we determined the need for a segmentation subnet or the replacement deep learning algorithm by logistic regression in screening clinical/radiological factors. Finally, an applet was constructed for future application.
RESULTS: The best model combined clinical/radiological factors and radiomic features. It achieved best discrimination (areas under the curve: 0·877 in the training dataset and 0·836 in the validation dataset), calibration, and decision curve. Its performance was not affected by the treatments and disease stages. The subgroups had statistical significance for time to macrovascular invasion (training: hazard ratio [HR] = 0·073, 95% confidence interval [CI]: 0·032-0·167, p < 0·001 and validation: HR = 0·090, 95%CI: 0·022-0·366, p < 0·001) and overall survival (training: HR = 0·344, 95%CI: 0·246-0·547, p < 0·001 and validation: HR = 0·489, 95%CI: 0·279 - 0·859, p = 0·003). Similar results were achieved when the patients were subdivided by the H-score. The subnet for segmentation and end-to-end deep learning algorithms improved the performance of the model.
CONCLUSIONS: Our multi-task deep learning network-based model successfully predicted future macrovascular invasion. In high-risk populations, besides the current first-line treatments, more therapies may be explored for macrovascular invasion.
摘要：
背景：构建预测肝细胞癌大血管浸润的模型以帮助及时干预。
方法：在2007年4月至2016年11月期间，回顾性地从五家中国医院收集了366例HCC病例：训练数据集包括来自四家医院的281例患者；外部验证数据集包括来自另一家医院的85例患者。构建了基于多任务深度学习网络的模型来预测未来的大血管侵犯。歧视,校准,和决策曲线进行比较，以确定最佳模型。我们使用最佳模型和相关图像异质性评分（H评分）比较了大血管浸润时间和总生存期。然后，我们通过逻辑回归确定了在筛查临床/放射学因素时需要分割子网或替代深度学习算法.最后,为将来的应用构建了一个小程序。
结果：结合临床/放射学因素和放射学特征的最佳模型。它实现了最佳的区分（曲线下的面积：训练数据集中的0·877和验证数据集中的0·836），校准,和决策曲线。其性能不受治疗和疾病阶段的影响。亚组对大血管浸润的时间具有统计学意义（训练：风险比[HR]=0·073，95％置信区间[CI]：0·032-0·167，p<0·001，验证：HR=0·090，95CI：0·022-0·366，p<0·001）和总体生存率（训练：HR=0·344，95CI：0·246-547，p=3当通过H评分对患者进行细分时，获得了类似的结果。分段和端到端深度学习算法的子网提高了模型的性能。
结论：我们基于多任务深度学习网络的模型成功预测了未来的大血管侵袭。在高危人群中,除了目前的一线治疗之外，对于大血管侵犯,可能会探索更多的治疗方法.
方法：在2007年4月至2016年11月期间，回顾性地从五家中国医院收集了366例HCC病例：训练数据集包括来自四家医院的281例患者；外部验证数据集包括来自另一家医院的85例患者。构建了基于多任务深度学习网络的模型来预测未来的大血管侵犯。歧视,校准,和决策曲线进行比较，以确定最佳模型。我们使用最佳模型和相关图像异质性评分（H评分）比较了大血管浸润时间和总生存期。然后，我们通过逻辑回归确定了在筛查临床/放射学因素时需要分割子网或替代深度学习算法.最后,为将来的应用构建了一个小程序。
结果：结合临床/放射学因素和放射学特征的最佳模型。它实现了最佳的区分（曲线下的面积：训练数据集中的0·877和验证数据集中的0·836），校准,和决策曲线。其性能不受治疗和疾病阶段的影响。亚组对大血管浸润的时间具有统计学意义（训练：风险比[HR]=0·073，95％置信区间[CI]：0·032-0·167，p<0·001，验证：HR=0·090，95CI：0·022-0·366，p<0·001）和总体生存率（训练：HR=0·344，95CI：0·246-547，p=3当通过H评分对患者进行细分时，获得了类似的结果。分段和端到端深度学习算法的子网提高了模型的性能。
结论：我们基于多任务深度学习网络的模型成功预测了未来的大血管侵袭。在高危人群中,除了目前的一线治疗之外，对于大血管侵犯,可能会探索更多的治疗方法.
