论文部分内容阅读
目的通过肝细胞癌影像特征和基因模块关联的基因影像学方法,获取生物可解释的影像学标记物。方法从癌症基因组数据库获取371例肝细胞癌基因表达数据,其中37例有对应的术前增强CT数据。从勾画的肿瘤区域提取639维定量影像特征,基于一致性指数标准,筛选出12个影像特征。对肝细胞癌基因表达数据进行聚类,得到聚类基因模块,并与影像特征构建Spearman相关图,筛选出与基因模块关联的影像特征,最后用Cox回归模型评估其是否具有预后功能。结果 12个影像特征中,4个与预后的基因模块显著相关,3个与生存预后显著相关。其中,小波分量高灰度级小区域增强特征,与代表糖链生物合成的基因模块显著相关,且该影像特征与病人的总体生存周期(P=0.006,风险比=0.16)显著相关;低灰度级长游程增强纹理特征,与代表血管生成的基因模块显著相关,且与总体生存周期(P=0.049,风险比=3.21)显著相关。结论描述肿瘤小波频率和纹理的影像特征有可解释的生物学含义,并且与生存周期显著相关,这些特征可作为肝细胞癌潜在的影像学标记物。
Objective To obtain bio-explainable imaging markers through the image features of hepatocellular carcinoma and the genetic imaging methods associated with gene modules. Methods 371 cases of hepatocellular carcinoma gene expression data were obtained from the cancer genome database, of which 37 cases had corresponding preoperative enhanced CT data. 639-dimensional quantitative image features were extracted from the delineated tumor regions, and 12 image features were screened based on the consensus index criteria. Hepatocellular carcinoma gene expression data were clustered to obtain a clustering gene module, and a Spearman correlation map was constructed with the image features to screen out image features associated with the gene module. Finally, a Cox regression model was used to evaluate whether it had prognostic function. Results Of the 12 image features, 4 were significantly associated with the prognostic gene module and 3 were significantly associated with survival prognosis. Among them, the wavelet component high grayscale small area enhancement features were significantly associated with the gene modules representing sugar chain biosynthesis, and this image feature was significantly associated with the patient’s overall survival period (P=0.006, hazard ratio=0.16); Long-term run length enhanced texture features and was significantly associated with gene modules representing angiogenesis, and was significantly associated with overall survival (P = 0.049, hazard ratio = 3.21). Conclusions The imaging characteristics of tumor wavelet frequency and texture have an interpretable biological meaning and are significantly associated with the survival period. These features can be used as potential imaging markers for hepatocellular carcinoma.