首页|期刊导航|磁共振成像|基于腹部CT体成分分析结合扩散峰度成像术前预测T3~T4期直肠癌肿瘤出芽级别的价值

基于腹部CT体成分分析结合扩散峰度成像术前预测T3~T4期直肠癌肿瘤出芽级别的价值OA

The value of preoperative prediction of tumor budding grade in stage T3-T4 rectal cancer based on abdominal CT body composition analysis combined with diffusion kurtosis imaging technique

中文摘要英文摘要

目的 探讨腹部CT体成分分析结合磁共振扩散峰度成像(diffusion kurtosis imaging,DKI)术前预测T3~T4期直肠癌肿瘤出芽(tumor budding,TB)级别的价值.材料与方法 回顾性分析89例术前行盆腔3.0 T MR及全腹CT检查并经手术病理证实的T3~T4期直肠腺癌患者资料,依据术后病理TB级别分为TB中低级别组(52例)、TB高级别组(37例).两名观察者分别测量两组病灶的DKI定量参数值,包括各向异性分数(fractional anisotropy,FA)值、平均扩散系数(mean diffusivity,MD)值、平均扩散峰度(mean kurtosis,MK)值.使用TotalSegmentator工具自动分割全腹CT L3水平皮下脂肪、内脏脂肪、肌肉.使用Python编程语言编写皮下脂肪面积(subcutaneous fat area,SFA)、内脏脂肪面积(visceral fat area,VFA)、肌肉面积(muscle area,MA)计算程序,并计算VFA与SFA比值(VFA/SFA)、SFA与标准体质量(standard body weight,SBW)比值(SFA/SBW)、VFA与SBW比值(VFA/SBW)、MA与SBW比值(MA/SBW)、VFA与SFA差值(VFA-SFA).使用组内相关系数(intra-class correlation coefficient,ICC)检验两名观察者间DKI各参数值测量一致性.采用独立样本t检验或Mann-Whitney U检验对组间DKI参数及体成分指标进行差异性比较.通过多因素logistic回归分析筛选出独立风险预测因子,并据此建立联合预测模型.利用受试者工作特征(receiver operating characteristic,ROC)曲线对各独立风险预测因子及联合预测模型的诊断效能进行评估.通过DeLong检验比较单一变量及联合预测模型的诊断效能差异,同时应用McNemar检验分析其敏感度及特异度差异.结果 DKI定量参数值在两名观察者间的测量一致性良好.TB高级别组MK、VFA、VFA/SBW值分别为0.938(0.891,1.019)、[150.620(107.685,199.840)]cm2、2.490(1.730,3.210),高于TB中低级别组 0.776(0.685,0.878)、[117.205(72.067,169.762)]cm2、2.115(1.242,2.650);TB高级别组MD值为(1.113±0.105)μm2/ms,低于TB中低级别组(1.347±0.257)μm2/ms;上述参数两组间差异均具有统计学意义(P<0.05).而两组在FA、SFA、MA、VFA/SFA、SFA/SBW、MA/SBW、VFA-SFA指标上差异未呈现出统计学意义(P>0.05).经多因素logistic回归模型分析,MK值[比值比:4.104,95%置信区间(confidence interval,CI):2.223~7.578]和VFA/SBW值[比值比:2.335(95%CI:1.300~4.195)]是预测TB级别的独立风险预测因子(P<0.05).MK、VFA/SBW值及两者的联合预测模型预测TB级别的曲线下面积(area under the curve,AUC)分别为0.838、0.627及0.886,敏感度分别为91.89%、72.97%及81.08%,特异度分别为67.31%、50.00%及80.77%.联合预测模型诊断效能优于VFA/SBW,且特异度优于MK、VFA/SBW,差异有统计学意义(P<0.05).结论 对于T3~T4期直肠癌患者,DKI可以有效地预测TB级别,体成分分析亦可初步对TB风险进行分层,两者结合可明显提高对TB级别预测的准确性,为术前预测TB风险级别提供了宿主与肿瘤局部结合更全面的信息.

Objective:This study aims to investigate the value of abdominal CT body composition analysis combined with diffusion kurtosis imaging(DKI)for the preoperative prediction of tumor budding(TB)grade in patients with T3 to T4 rectal cancer.Materials and Methods:A retrospective analysis was conducted on data from 89 patients diagnosed with T3 to T4 rectal adenocarcinoma,confirmed by postoperative pathology.All patients underwent pelvic 3.0 T MRI and whole abdominal CT examinations prior to surgery.Based on the pathological results,the TB grade was categorized into two groups:a low-intermediate grade TB group(52 cases)and a high grade TB group(37 cases).Two observers measured the quantitative parameters of DKI in the lesions of both groups,including fractional anisotropy(FA)value,mean diffusivity(MD)value,and mean kurtosis(MK)value.The TotalSegmentator tool was utilized to automatically segment subcutaneous fat,visceral fat,and muscle at the L3 level of the abdominal CT scans.Python software facilitated the automatic calculation of subcutaneous fat area(SFA),visceral fat area(VFA),and muscle area(MA).The ratios of VFA to SFA(VFA/SFA),SFA to standard body weight(SBW)(SFA/SBW),VFA to SBW(VFA/SBW),MA to SBW(MA/SBW),and the difference between VFA and SFA(VFA-SFA)were calculated.The consistency of DKI parameters assessed by two observers was evaluated using the intra-class correlation coefficient(ICC).To examine the differences in DKI parameters and body composition indicators across the groups,either a Mann-Whitney U test or an independent sample t-test was utilized.A multivariate logistic regression analysis was performed to determine independent risk factors,which facilitated the creation of a combined prediction model based on these factors.The diagnostic performance of each individual risk factor,as well as the combined prediction model,was evaluated through the receiver operating characteristic(ROC)curve.To compare the diagnostic efficacy between the individual variables and the combined model,the DeLong test was employed,whereas the McNemar test was used to assess differences in sensitivity and specificity.Results:The consistency of the DKI quantitative parameters measured by the two observers was good,with ICC values exceeding 0.75.In the high grade TB group,the MK,VFA,and VFA/SBW values were 0.938(0.891,1.019),[150.620(107.685,199.840)]cm2,and 2.490(1.730,3.210),respectively,significantly surpassing those in the low-intermediate grade TB group,which had values of 0.776(0.685,0.878),[117.205(72.067,169.762)]cm2,and 2.115(1.242,2.650).The MD value in the high grade TB group was(1.113±0.105)μm2/ms,which was lower than that in the low-intermediate grade TB group(1.347±0.257)μm²/ms.The differences between the two groups for all the aforementioned parameters were statistically significant(P<0.05).The results of multivariate logistic regression analysis indicated that the MK value[odds ratio:4.104,95%confidence interval(CI):2.223 to 7.578]and the VFA/SBW value[odds ratio:2.335(95%CI:1.300 to 4.195)]were independent risk predictors for TB grade(P<0.05).The area under the curve(AUC)for predicting TB grade based on the MK and VFA/SBW values,as well as their combined prediction model,were 0.838,0.627,and 0.886,respectively.The sensitivity of these models was 91.89%,72.97%,and 81.08%,respectively,while the specificity was 67.31%,50.00%,and 80.77%,respectively.The diagnostic efficacy of the combined prediction model was superior to that of VFA/SBW alone,and its specificity exceeded that of both MK and VFA/SBW,with statistically significant differences(P<0.05).Conclusions:For patients with T3 to T4 rectal cancer,DKI has been shown to effectively predict TB grade.Additionally,body composition analysis serves to preliminarily stratify the risk of TB.The integration of these two modalities significantly enhances the accuracy of predicting TB grade,thereby providing more comprehensive insights into the interplay between host factors and local tumor characteristics for preoperative TB risk assessment.

陈安良;张钦和;李文豪;谢素玲;王秀林;王洪凯;孙祎楠;魏强;刘爱连

大连医科大学附属第一医院放射科,大连 116011||大连市医学影像人工智能工程技术研究中心,大连 116011||辽宁省超极化磁共振专业技术创新中心,大连 116011大连医科大学附属第一医院放射科,大连 116011||大连市医学影像人工智能工程技术研究中心,大连 116011||辽宁省超极化磁共振专业技术创新中心,大连 116011大连医科大学附属第一医院放射科,大连 116011大连医科大学附属第一医院病理科,大连 116011大连医科大学附属第一医院干细胞临床研究机构,大连 116011大连市医学影像人工智能工程技术研究中心,大连 116011||大连理工大学医学部,大连 116011大连理工大学医学部,大连 116011大连医科大学附属第一医院放射科,大连 116011大连医科大学附属第一医院放射科,大连 116011||大连市医学影像人工智能工程技术研究中心,大连 116011||辽宁省超极化磁共振专业技术创新中心,大连 116011

医药卫生

直肠癌肿瘤出芽体成分分析定量计算机断层扫描磁共振成像扩散峰度成像

rectal cancertumor buddingbody composition analysisquantitative computed tomographymagnetic resonance imagingdiffusion kurtosis imaging

《磁共振成像》 2026 (9)

16-24,9

辽宁省科技计划联合计划项目(编号:2025-MSLH-199)大连医科大学附属第一医院院内基金项目(编号:2019HZ007)Joint Research Program of Liaoning Provincial Science and Technology Plan(No.2025-MSLH-199),Fund Project of the First Affiliated Hospital of Dalian Medical University(No.2019HZ007).

10.12015/issn.1674-8034.2026.09.003

评论