目的:探讨基于钆塞酸二钠增强MRI肝胆期影像组学在肝细胞癌(HCC)病理分化程度中的预测价值。方法:回顾性收集145例经术后病理证实为HCC患者资料,根据病理分级将患者分为低危险组(100例)和高危险组(45例),前者包括高分化(11例)、中-高分...目的:探讨基于钆塞酸二钠增强MRI肝胆期影像组学在肝细胞癌(HCC)病理分化程度中的预测价值。方法:回顾性收集145例经术后病理证实为HCC患者资料,根据病理分级将患者分为低危险组(100例)和高危险组(45例),前者包括高分化(11例)、中-高分化(11例)、中分化(78例)HCC,后者包括低分化(21例)、中-低分化(24例)HCC。所有患者术前2周均行钆塞酸二钠增强MRI检查。通过分析得出对HCC有影响的临床及传统影像学特征,建立临床-影像模型;利用ITK⁃SNAP软件在肝胆特异期图像上手动勾画肿瘤感兴趣区(ROIs),利用FAE(FeAture Explorer V.0.5.5)软件提取特征,使用最大相关最小冗余法对特征降维,通过建立XGBoost二分类模型来预测诊断效能;最终将临床-影像模型与影像组学模型相结合,建立联合模型。采用受试者工作特征(ROC)曲线下面积、准确度、灵敏度、特异度等指标来评价各模型诊断效能。结果:测试组中临床-影像模型、影像组学模型以及联合模型的ROC曲线下面积值分别为0.66、0.70、0.76,准确度分别为63.6%、63.6%、77.3%,灵敏度分别为71.4%、85.7%、57.1%,特异度分别为60.0%、53.3%、86.7%。结论:基于钆塞酸二钠增强MRI肝胆期影像组学的模型在预测HCC病理分化中有一定的价值。展开更多
Objective To explore the value of contrast-enhanced ultrasound(CEUS)for predicting differentiation degree of hepatocellular carcinoma(HCC).Methods Totally 86 HCC patients confirmed by postoperative pathology were retr...Objective To explore the value of contrast-enhanced ultrasound(CEUS)for predicting differentiation degree of hepatocellular carcinoma(HCC).Methods Totally 86 HCC patients confirmed by postoperative pathology were retrospectively enrolled and divided into poorly differentiated,moderately differentiated and highly differentiated groups according to postoperative Edmondson-Steiner grading.Preoperative CEUS parameters were compared among groups,and binary logistic regression was used to analyze CEUS-related independent predictors of HCC with different differentiation.The receiver operating characteristic curves of parameters being significant different among groups were drawn,the areas under the curve(AUC)were calculated,and the efficacy for predicting HCC with different differentiation degree was evaluated.Results There were 29 cases in poorly differentiated group,37 in moderately differentiated group and 20 cases in highly differentiated group.The arrival time of contrast agent in poorly differentiated group was earlier than that in moderately and high differentiated groups(both P<0.05),while in moderately differentiated group was not significantly different with that in highly differentiated group(P>0.05).The washout grade were significantly different between each 2 groups(all P<0.05).The arrival time of contrast agent and washout grade were independent predictors of highly or poorly differentiated,moderately or poorly differentiated,moderate-highly or poorly differentiated HCC,and washout grade also was independent predictor of highly or moderately differentiated HCC(all P<0.05).The AUC of the arrival time of contrast agent for predicting highly or moderately differentiated,highly or poorly differentiated,moderately or poorly differentiated,moderate-highly or poorly differentiated HCC was 0.615,0.787,0.690 and 0.724,respectively,while of washout grade was 0.801,0.927,0.795 and 0.841,respectively.Conclusion CEUS could be used to effectively predict differentiation degree of HCC.展开更多
文摘目的:探讨基于钆塞酸二钠增强MRI肝胆期影像组学在肝细胞癌(HCC)病理分化程度中的预测价值。方法:回顾性收集145例经术后病理证实为HCC患者资料,根据病理分级将患者分为低危险组(100例)和高危险组(45例),前者包括高分化(11例)、中-高分化(11例)、中分化(78例)HCC,后者包括低分化(21例)、中-低分化(24例)HCC。所有患者术前2周均行钆塞酸二钠增强MRI检查。通过分析得出对HCC有影响的临床及传统影像学特征,建立临床-影像模型;利用ITK⁃SNAP软件在肝胆特异期图像上手动勾画肿瘤感兴趣区(ROIs),利用FAE(FeAture Explorer V.0.5.5)软件提取特征,使用最大相关最小冗余法对特征降维,通过建立XGBoost二分类模型来预测诊断效能;最终将临床-影像模型与影像组学模型相结合,建立联合模型。采用受试者工作特征(ROC)曲线下面积、准确度、灵敏度、特异度等指标来评价各模型诊断效能。结果:测试组中临床-影像模型、影像组学模型以及联合模型的ROC曲线下面积值分别为0.66、0.70、0.76,准确度分别为63.6%、63.6%、77.3%,灵敏度分别为71.4%、85.7%、57.1%,特异度分别为60.0%、53.3%、86.7%。结论:基于钆塞酸二钠增强MRI肝胆期影像组学的模型在预测HCC病理分化中有一定的价值。
文摘Objective To explore the value of contrast-enhanced ultrasound(CEUS)for predicting differentiation degree of hepatocellular carcinoma(HCC).Methods Totally 86 HCC patients confirmed by postoperative pathology were retrospectively enrolled and divided into poorly differentiated,moderately differentiated and highly differentiated groups according to postoperative Edmondson-Steiner grading.Preoperative CEUS parameters were compared among groups,and binary logistic regression was used to analyze CEUS-related independent predictors of HCC with different differentiation.The receiver operating characteristic curves of parameters being significant different among groups were drawn,the areas under the curve(AUC)were calculated,and the efficacy for predicting HCC with different differentiation degree was evaluated.Results There were 29 cases in poorly differentiated group,37 in moderately differentiated group and 20 cases in highly differentiated group.The arrival time of contrast agent in poorly differentiated group was earlier than that in moderately and high differentiated groups(both P<0.05),while in moderately differentiated group was not significantly different with that in highly differentiated group(P>0.05).The washout grade were significantly different between each 2 groups(all P<0.05).The arrival time of contrast agent and washout grade were independent predictors of highly or poorly differentiated,moderately or poorly differentiated,moderate-highly or poorly differentiated HCC,and washout grade also was independent predictor of highly or moderately differentiated HCC(all P<0.05).The AUC of the arrival time of contrast agent for predicting highly or moderately differentiated,highly or poorly differentiated,moderately or poorly differentiated,moderate-highly or poorly differentiated HCC was 0.615,0.787,0.690 and 0.724,respectively,while of washout grade was 0.801,0.927,0.795 and 0.841,respectively.Conclusion CEUS could be used to effectively predict differentiation degree of HCC.
