基于双能CT虚拟单能量技术对非小细胞肺癌影像组学特征变化的研究

Changes of radiomic feature of non-small cell lung cancer based on dual-energy CT virtual monoenergetic technology

目的探讨双能CT(DECT)不同能级下虚拟单能量成像(VMI)对非小细胞肺癌(NSCLC)肿瘤靶区影像组学特征的影响。方法回顾性分析2020-01-01-2022-01-31山东省肿瘤医院82例行光谱CT增强扫描的NSCLC患者病例资料,其中男50例,女32例。分别在动脉期120 kVp混合能量图像和40、60、70、80、100、120 keV不同能级VMI下勾画NSCLC的大体肿瘤区(GTV),提取影像组学特征。以每个特征的百分变异系数(%COV)评估特征的稳定性,以类内一致性系数(ICC)评估每个特征的一致性,计算每个特征在不同能级梯度(40~60、60~80、80~100、100~120 keV)下的平均变化率。结果共提取93个影像组学特征,120 kVp图像及不同能级VMI之间的特征稳定性差异无统计学意义,F=0.324,P=0.925。ICC结果表明,除了一阶的5个特征外,其他特征在不同能级VMI之间具有较好的一致性(ICC≥0.75)。120 kVp图像与70 keV VMI的特征一致性最高(ICC=0.98±0.03),与40 keV VMI的特征一致性最低(ICC=0.85±0.19)。40 keV组与120kVp图像的特征一致性与其他能级组与120kVp图像特征一致性相比,差异有统计学意义,F=26.165,P<0.001。Spearman秩次相关分析表明,特征变化率与不同能级梯度之间呈中度负相关性,r=-0.465,P<0.001,特征值变化率的绝对值在相同区间内随能级的递增逐渐下降。结论不同能级VMI对影像组学特征稳定性的影响无统计学差异;常规混合能量图像与VMI影像组学特征的一致性存在较大差异,影像组学特征对于能量变化的灵敏度不同。基于DECT VMI的影像组学特征可能提供更多有价值的成像信息。

Objective To investigate the effect of virtual monoenergetic images(VMI)at different energy levels of dual-energy computed tomography(DECT)on the radiomic features of tumor targets in non-small cell lung cancer(NSCLC).Methods Totally 82patients with NSCLC who underwent spectral CT enhancement scanning in Shandong Cancer Hospital from January 1,2020to January 31,2022were analyzed retrospectively,including 50males and 32females.The gross tumor volume(GTV)of NSCLC was delineated in arterial phase 120kVp mixed energy images and different energy levels of 40,60,70,80,100,120keV VMI,respectively,and the radiomic features were extracted.The coefficient of variation(%COV)of each feature was calculated to evaluate the stability of each feature,and the intra-class correlation coefficient(ICC)was used to evaluate the consistency of each feature.The average change rate of each feature was calculated under different energy level gradients(40-60,60-80,80-100,100-120keV).Results A total of 93radiomic features were extracted,and there was no significant difference in feature stability between 120kVp images and different energy levels of VMI(F=0.324,P=0.925).ICC results showed that except for the five features of first-order,other features had good consistency among different energy levels of VMI(ICC≥0.75).The feature consistency between 120kVp images and 70keV VMI was the highest(ICC=0.98±0.03),and that with 40keV VMI was the lowest(ICC=0.85±0.19).The difference between the feature consistency of 40keV group and 120kVp image and that of other energy level groups and 120kVp image was statistically significant(F=26.165,P<0.001).Spearman rank correlation analysis showed that there was a moderate negative correlation between the feature change rate and the gradients of different energy levels(r=-0.465,P<0.001).The absolute value of the feature change rate decreased gradually with the increase of energy levels in the same interval.Conclusionuc2 There is no significant difference in the effect of different energy levels of VMI on the stability of radiomic features,but there is a significant difference in the consistency between conventional mixed energy images and VMI radiomic features.The sensitivity of each feature to the change of energy is different.The radiomic features based on DECT VMI may provide more valuable imaging information.