文章摘要
洪敏萍,周长玉,卜阳阳,等.基于多参数MRI影像组学特征预测肉芽肿性乳腺炎中医证型的研究[J].浙江中医药大学学报,2022,(5):483-489.
基于多参数MRI影像组学特征预测肉芽肿性乳腺炎中医证型的研究
Predicting TCM Syndrome of Granulomatous Mastitis Based on Multi-parameter MRI Radiomic Features
DOI:10.16466/j.issn1005-5509.2022.05.002
中文关键词: 乳腺  影像组学  肉芽肿性乳腺炎  中医分型  多参数磁共振成像  预测模型  血瘀证  气血不足证
英文关键词: breast  radiomics  granulomatous mastitis  TCM syndrome types  multi-parameter MRI  prediction model  blood stasis  deficiency of Qi and blood
基金项目:浙江省中医药科技计划项目(2021ZB089);浙江省中医药科技计划项目(2018ZA037);浙江省医药卫生科研项目(2021KY224)
作者单位
洪敏萍 浙江中医药大学附属嘉兴中医医院 浙江嘉兴 314000
浙江中医药大学附属第一医院
浙江中医药大学第一临床医学院 
周长玉 浙江中医药大学附属第一医院
浙江中医药大学第一临床医学院 
卜阳阳 浙江中医药大学附属第一医院
浙江中医药大学第一临床医学院 
许茂盛 浙江中医药大学附属第一医院
浙江中医药大学第一临床医学院 
赵虹 浙江中医药大学附属第一医院
浙江中医药大学第一临床医学院 
摘要点击次数: 153
全文下载次数: 138
中文摘要:
      [目的] 探讨多参数磁共振成像(magnetic resonance imaging,MRI)影像组学特征构建的模型在预测肉芽肿性乳腺炎(granulomatous mastitis,GM)中医辨证分型中的价值。[方法] 回顾性分析经病理证实的GM病例111例,所有入组病例根据中医诊断标准分为标阳本阴兼血瘀证和标阳本阴兼气血不足证,其中标阳本阴兼血瘀证58例,标阳本阴兼气血不足53例,采用随机抽样方法将病例按7:3分为训练组及验证组,通过深睿科研平台软件,在动态增强磁共振成像(dynamic contrast-enhanced magnetic resonance imaging,DCE MRI)第2期、弥散加权成像(diffusion-weighted imaging,DWI)、表观弥散系数(apparent diffusion coefficient,ADC)图像中各提取了1 648个影像组学特征;通过随机森林算法及LASSO回归分析建立4个模型:DCE模型、ADC模型、DWI模型及联合模型(DCE+ADC+DWI),通过受试者工作特征(receiver operating characteristic,ROC)曲线下面积(area under curve,AUC)、特异度、敏感度、准确度评估预测性能。 [结果] 4个影像组学模型(DCE模型、ADC模型、DWI模型及联合模型)在预测GM中医分型的训练组中的AUC分别为0.84、0.82、0.81、0.92,验证组中的AUC分别为0.79、0.73、0.70、0.82,联合模型诊断效能最高,验证组中的准确度、敏感度、特异度分别为0.82、0.94、0.67。 [结论] 基于多参数MRI的影像组学模型具有预测GM中医分型的潜力,联合DCE、DWI、ADC序列比单一序列更佳,本研究结果可为中医辨证施治提供客观化、标准化的依据。
英文摘要:
      [Objective] To explore the value of the model constructed by multi-parameter magnetic resonance imaging(MRI) radiomic features in predicting traditional Chinese medicine(TCM) syndrome of granulomatous mastitis(GM). [Methods] According to the diagnostic criteria of TCM, all GM patients were divided into two groups: 58 patients of superficial Yang and root Yin with blood stasis syndrome and 53 patients of superficial Yang and root Yin with Qi and blood deficiency syndrome. According to the method of random sampling, the patients were divided into training group and verification group at 7:3. Based on 1 648 imaging features of dynamic contrast-enhanced magnetic resonance imaging(DCEMRI) phase at the second scan images, diffusion-weighted imaging(DWI) and apparent diffusion coefficient(ADC) sequence, four models were established by random forest algorithm and LASSO regression analysis: DCE model, ADC model, DWI model and combined model(DCE+ADC+DWI). The prediction performance was evaluated by the area under curve(AUC) of receiver operating characteristic(ROC), accuracy, specificity and sensitivity. [Results] The AUC of the four imaging models in training group for predicting the TCM classification of GM was 0.84, 0.82, 0.81, 0.92 respectively, and the AUC of verification group was 0.79, 0.73, 0.70, 0.82, respectively. The specificity, sensitivity and accuracy of verification group were 0.82, 0.94 and 0.67, respectively. [Conclusion] The imaging model based on multi-parameter MRI has the potential to predict the TCM classification of GM. The combination of DCE, DWI and ADC sequence is better than a single sequence, which can provide an objective and standardized basis for TCM syndrome differentiation and treatment.
查看全文   查看/发表评论  下载PDF阅读器
关闭