基于CT影像组学评估射频消融术治疗晚期肝癌的效果

CT-based imaging radiomics model in assessment of efficacy of radiofrequency ablation for advanced hepatocellular carcinoma

目的探讨基于CT影像组学评估射频消融术(RFA)治疗晚期肝癌的临床效果。方法观察性研究。抽取2020年1月至2023年1月鹤壁市人民医院收治的晚期肝癌患者63例,均行RFA治疗,根据相关标准判定疗效(缓解、未缓解)。提取63例患者RFA术前CT检查图像,挑选感兴趣区后再分割图像中病灶,获得纹理特征;按照7∶3比例分组,训练组44例(缓解31例,未缓解13例),验证组19例(缓解13例,未缓解6例)。使用套索方式予以降维,构建相应的数据模型;采用晚期肝癌预测模型决策曲线以及受试者工作特征曲线下面积(AUC),分析影像组学模型对RFA疗效的预测价值。结果训练组、验证组影像组学评分缓解者低于未缓解者,P<0.05。基于CT图像提取12个纹理特征,建立相应影像组学模型结果显示,训练组AUC为0.844,灵敏度和特异度为70.62%、80.54%;验证组AUC为0.726,灵敏度和特异度为72.58%、82.39%。影像组学预测模型的决策曲线在0.17~0.75阈值内。结论晚期肝癌患者CT影像组学能有效预测RFA的疗效,有助于指导临床,实现个性化诊断和治疗。

Objective To investigate that CT-based imaging radiomics model in assessment of the efficacy of radiofrequency ablation(RFA)for advanced hepatocellular carcinoma.Methods A total of 63 patients with advanced hepatocellular carcinoma admitted to People’s Hospital of Hebi from January 2020 to January 2023 were selected for the observational trail.All of them were treated by RFA,and the treatment efficacy(remission and non-remission)were evaluated by relative criteria.The CT images before RFA of the 63 patients were collected,and the lesions in the images were segmented to obtain texture features after the regions of interest were selected.The 63 patients were grouped with the ratio of 7∶3;44 cases were enrolled in the training group,including 31 cases with remission and 13 cases without remission.Nineteen cases were allocated into the validation group,including 13 cases with remission and 6 cases without remission.The lasso method was used to reduce the dimension,and the corresponding data model was constructed.The decision curve of the prediction model and the area under the receiver operating characteristic curve(AUC)of the prediction model of advanced liver cancer were used to analyze the predictive value of the imaging radiomics model on the efficacy of RFA.Results The radiomics scores of the patients with remission were lower than those of patients without remission in the training group and the validation group(P<0.05).According to the model constructed on 12 texture features extracted from CT images,the AUC of the training group was 0.844,and the sensitivity and specificity were 70.62%and 80.54%,respectively.The AUC of the validation group was 0.726,and the sensitivity and specificity were 72.58%and 82.39%,respectively.The decision curve of the radiomics prediction model was within the threshold of 0.17-0.75.Conclusions CT imaging radiomics model can effectively predict and evaluate the efficacy of RFA for advanced hepatocellular carcinoma,which is useful in guiding clinical practice,personalized diagnosis and treatment for patients.