IVIM-DWI联合基于T2WI纹理分析术前鉴别直肠癌T2与T3分期的应用价值

目的 探讨体素内不相干运动扩散加权成像（IVIM-DWI）和基于 T₂WI 纹理分析术前鉴别直肠癌 T₂与 T₃分期的价值。 方法 回顾性分析 2016 年 6 月至 2023 年 5 月于中国科学技术大学附属第一医院西区（安徽省肿瘤医院）就诊且经手术病理证实的 105 例直肠癌患者临床资料。根据术后病理 T 分期将患者分为 T₂期组（n=57）和 T₃期组（n=48）。所有患者均于术前行常规 MRI 及IVIM-DWI 检查。测量患者癌灶的 IVIM-DWI 参数表观扩散系数（ADC）、D、D*和 f，并提取 T₂WI 序列的三维纹理特征。比较两组患者IVIM-DWI 参数和纹理参数的差异，建立纹理特征模型、IVIM-DWI 和纹理特征参数的联合模型。用受试者工作特征（ROC）曲线评估各参数和模型术前对直肠癌 T₂和 T₃分期的鉴别效能。 结果 IVIM-DWI 参数中，T₂期组 D 值大于 T₃期组 D 值，T₂期组 f 值小于 T₃期组f 值，差异均有统计学意义（P＜0.05），ADC 值和 D* 值的差异无统计学意义（P＞0.05）。从 T₂WI 提取的 1 874 个纹理参数中，筛选出 4个具有统计学意义的纹理参数：firstorder_90Percentile、glrlm_LRLGLE、glszm_SZN 和 glszm_SAHGLE。D 值、f 值及 4 个纹理特征参数的曲线下面积（AUC）分别为 0.705，0.664，0.840，0.740，0.801，0.656。纹理特征模型的 AUC 为 0.952，联合模型的 AUC 为 0.956。联合模型的 AUC 大于 D 值、f 值及 4 个纹理特征参数的 AUC，其灵敏度为 93.9%，特异度为 87.5%。 结论 IVIM-DWI 联合基于 T₂WI 纹理分析对术前直肠癌的 T₂期和 T₃期具有良好的鉴别效能，一定程度上弥补常规 MRI 技术之不足。

Value of IVIM-DWI combined with texture analysis based on T2WI to differentiate T2 from T3 stage in rectal cancer preoperatively

Objective To investigate the value of IVIM-DWI combined with texture analysis based on T₂WI in differentiating T₂ from T 3 stage of rectal cancer preoperatively. Methods Retrospective analysis was performed on the clinical data of 105 cases of rectal cancer confirmed by surgical pathology in West Branch of the First Affiliated Hospital of University of Science and Technology of China (Anhui Provincial Cancer Hospital) from June 2016 to May 2023. The enrolled patients were divided into the T₂ stage group (n=57) and the T₃ stage group (n=48) according to T stage. All patients underwent conventional rectal MRI and IVIM-DWI scan before surgery. IVIM-DWI parameters of primary rectal cancer were measured, including ADC, D, D*and f, and the three-dimensional texture features based on T₂WI were extracted. The differences in IVIM-DWI parameters and texture feature parameters were compared between the two groups, then the texture feature model and combined model based on IVIM-DWI and texture feature parameters were established. The receiver operating characteristic (ROC) curves were used to evaluate the performance of each parameter and model in the preoperative prediction about T₂ and T₃ stage in rectal cancer. Results Among all the IVIM-DWI parameters, the D value of T₂ stage group was bigger than T₃ stage group, and the f value of T₂ stage group was smaller than T₃ stage group. Both of them were significantly different (P＜0.05), and no statistical significance was found in the ADC and D* values between two groups (both P＞0.05). Four texture parameters with statistical significance, including firstorder_90Percentile, glrlm_LRLGLE, glszm_SZN and glszm_SAHGLE, were selected totally from 1874 texture parameters, which extracted from T₂WI sequence. The AUC of D, f and four texture feature parameters was 0.705, 0.664, 0.840, 0.740, 0.801, 0.656 respectively. The AUC of texture feature model and the combined model was 0.952 and 0.956, respectively. The AUC of combined model was larger than D, f and the four texture feature parameters (P<0.05), and the sensitivity was 93.9% and the specificity was 87.5%.Conclusion IVIM-DWI parameters combined with texture analysis based on T 2WI show better efficacy in differentiating T₂ from T₃ stage of rectal cancer preoperatively，which may remedy the weakness of conventional MRI to some extent.