
浏览全部资源
扫码关注微信
1. 中南大学湘雅医学院附属肿瘤医院,湖南省肿瘤医院放射诊断科,湖南,长沙,410006
2. 中南大学湘雅医学院附属肿瘤医院,湖南省肿瘤医院结直肠外科,湖南,长沙,410006
3. 中南大学湘雅医学院附属肿瘤医院,湖南省肿瘤医院放疗科,湖南,长沙,410006
4. 中南大学湘雅医学院附属肿瘤医院,湖南省肿瘤医院病理科,湖南,长沙,410006
网络出版:2020-04-28,
纸质出版:2020-04-28
移动端阅览
刘思野,文露,侯静,等. 基线动态对比增强磁共振成像纹理特征对直肠癌放化疗病理反应状态的预测价值[J]. 肿瘤影像学, 2020, 29(2): 148-154 https://doi.
org/10.19732/j.cnki.2096-6210.2020.02.013
刘思野,文露,侯静,等. 基线动态对比增强磁共振成像纹理特征对直肠癌放化疗病理反应状态的预测价值[J]. 肿瘤影像学, 2020, 29(2): 148-154 https://doi. DOI: 10.19732/j.cnki.2096-6210.2020.02.013.
org/10.19732/j.cnki.2096-6210.2020.02.013 DOI:
目的:
探讨基线动态对比增强磁共振成像(dynamic contrast-enhanced magnetic resonance imaging,DCE-MRI)参数图纹理特征在预测直肠癌原发灶新辅助放化疗病理反应方面的价值。
方法:
搜集38例局部进展期的直肠癌患者治疗前的DCE-MRI及术后病理学检查资料,采用受试者工作特征(receiver operating characteristic,ROC)曲线比较治疗前DCE-MRI参数[容量转运常数(volume transfer constant,K
trans
)、速率常数(rate constant,K
ep
)、血浆容量分数(plasma volume fraction,fPV)、血管外细胞外间隙容量分数(extravascular extracellular volume fraction,V
e
)]及参数图纹理特征对原发灶病理学完全缓解(pathological complete response,pCR)的预测效能。
结果:
基线DCE-MRI参数值在pCR组与非pCR组间差异无统计学意义。与非pCR组相比,pCR组的偏度 V
e
(Skewness V
e
)、相关度 K
ep
(Correlat K
ep
)、差分方差
fPV
(difference variance of
fPV
,DifVarnc
fPV
)、差分方差 K
ep
(difference variance of K
ep
,DifVarnc K
ep
)、差分熵
fPV
(difference entropy of
fPV
,DifEntrp
fPV
)、差分熵 V
e
(difference entropy of V
e
,DifEntrp V
e
)、熵
fPV
(Entropy
fPV
)、熵 Ve (Entropy
Ve
)、和平均数
fPV
(sum average of
fPV,SumAverg
fPV
)、和熵
fPV
(sum entropy of fPV,SumEntrp
fPV
)及和熵 V
e
(sum entropy of V
e
,SumEntrp
Ve
)值更高,均数 V
e
(Mean V
e
)、角二阶矩
fPV
(angular second moment of fPV,AngScMom
fPV
)、角二阶矩 Ve(angular second moment of V
e
,AngScMom
Ve
)、对比度
ktrans
(Contrast ktrans )、逆差矩
fPV
(Inverse Difference Moment of fPV,InvDfMom
fPV
)、和平均数K
trans
(sum average of K
trans
,SumAverg
ktrans
)及和平均数 V
e
(sum average of V
e
,SumAverg
Ve
)值更低。单个纹理特征预测pCR的ROC曲线的曲线下面积(area under curve,AUC)值为0.571~0.817;在多变量分析中,基于二级纹理特征构建的logistic回归模型预测pCR的AUC高于基于一级纹理特征构建的logistic回归模型(
P
=0.028)。
结论:
基线DCE-MRI参数图纹理特征可能在预测直肠癌原发灶新辅助放化疗病理反应状态方面有潜在价值。
Objective:
To evaluate the utility of baseline texture features based on dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in predicting the pathological complete response (pCR) to neoadjuvant chemoradiotherapy in locally advanced rectal cancer (LARC).
Methods:
Thirty-eight patients with LARC received DCE-MRI before neoadjuvant chemoradiotherapy. The data of DCE-MRI and the postoperative pathological data were collected. Receiver operating characteristic (ROC) curves were used to determine the efficiency of the DCE-MRI parameters (K
trans
K
ep
fPV and V
e
) and texture features derived from the DCE-MRI maps (K
trans
K
ep
fPV and V
e
) for identifying pCR.
Results:
The DCE-MRI parameters (K
trans
K
ep
fPV and V
e
) showed no significant difference between the pCR group and the non-pCR group. Compared to the non-pCR group
the pCR group exhibited significantly higher values of Skewne
ss
Ve
Correlat
Kep
difference variance of fPV (DifVarnc
fPV
)
difference variance of K
ep
(DifVarnc
Kep
)
difference entropy of fPV (DifEntrp
fPV
)
difference entropy of V
e
(DifEntrp
Ve
)
Entropy
fPV
Entropy
Ve
sum average of fPV (SumAverg
fPV
)
sum entropy of fPV (SumEntrp
fPV
) and sum entropy of V
e
(SumEntrp
Ve
)
as well as obviously lower values of Mean
Ve
angular second moment of fPV (AngScMom
fPV
)
angular second moment of V
e
(AngScMom
Ve
)
Contrast
ktrans
inverse difference moment of fPV (InvDfMom
fPV
)
sum average of K
trans
(SumAverg
Ktrans
) and sum average of V
e
(SumAverg
Ve
). In univariate analysis
the area under the ROC curve (AUC) values for individual predictor to identify pCR ranged from 0.571~0.817. In multivariate analysis
the gray level co-occurrence matrix texture features-based (GLCM TFs-based) logistic regression model had significantly higher AUC value than that based on the first-order TFs (
P
=0.028) in the differentiation between pCR and non-pCR.
Conclusion:
TFs based on baseline DCE-MRI may be potential to predict the pathological response of LARC receiving neoadjuvant chemoradiotherapy.
0
浏览量
1654
下载量
6
CNKI被引量
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621