磁共振成像的放射组学分析预测鼻咽癌患者诱导化疗的反应。
Radiomic analysis of MRI for prediction of response to induction chemotherapy in nasopharyngeal carcinoma patients.
机构信息
Department of Radiology, Sichuan Clinical Research Center for Cancer, Sichuan Cancer Hospital & Institute, Sichuan Cancer Center, Affiliated Cancer Hospital of University of Electronic Science and Technology of China, Chengdu, China.
Department of Radiology, Sichuan Clinical Research Center for Cancer, Sichuan Cancer Hospital & Institute, Sichuan Cancer Center, Affiliated Cancer Hospital of University of Electronic Science and Technology of China, Chengdu, China.
出版信息
Clin Radiol. 2023 Sep;78(9):e644-e653. doi: 10.1016/j.crad.2023.05.012. Epub 2023 Jun 7.
AIM
To establish and validate radiomic models for response prediction to induction chemotherapy (IC) in nasopharyngeal carcinoma (NPC) using the radiomic features from pretreatment MRI.
MATERIALS AND METHODS
This retrospective analysis included 184 consecutive NPC patients, 132 in the primary cohort and 52 in the validation cohort. Radiomic features were derived from contrast-enhanced T1-weighted imaging (CE-T1) and T2-weighted imaging (T2-WI) for each subject. The radiomic features were then selected and combined with clinical characteristics to build radiomic models. The potential of the radiomic models was evaluated based on its discrimination and calibration. To measure the performance of these radiomic models in predicting the treatment response to IC in NPC, the area under the receiver operating characteristic curve (AUC), and sensitivity, specificity, and accuracy were used.
RESULTS
Four radiomic models were constructed in the present study including the radiomic signature of CE-T1, T2-WI, CE-T1 + T2-WI, and the radiomic nomogram of CE-T1. The radiomic signature of CE-T1 + T2-WI performed well in distinguishing response and non-response to IC in patients with NPC, which yielded an AUC of 0.940 (95% CI, 0.885-0.974), sensitivity of 83.1%, specificity of 91.8%, and accuracy of 87.1% in the primary cohort, and AUC of 0.952 (95% CI, 0.855-0.992), sensitivity of 74.2%, specificity of 95.2%, and accuracy of 82.7% in the validation cohort.
CONCLUSION
MRI-based radiomic models could be helpful for personalised risk stratification and treatment in NPC patients receiving IC.
目的
利用治疗前磁共振成像（MRI）的放射组学特征，建立并验证预测鼻咽癌（NPC）诱导化疗（IC）反应的放射组学模型。
材料与方法
本回顾性分析纳入了 184 例连续的 NPC 患者，其中 132 例来自于主要队列，52 例来自于验证队列。为每位患者提取对比增强 T1 加权成像（CE-T1）和 T2 加权成像（T2-WI）的放射组学特征。然后，选择放射组学特征并与临床特征相结合，构建放射组学模型。基于其判别和校准能力，评估放射组学模型的预测能力。为了评估这些放射组学模型在预测 NPC 患者 IC 治疗反应中的性能，使用受试者工作特征曲线（ROC）下面积（AUC）、敏感性、特异性和准确性来衡量。
结果
本研究构建了 4 个放射组学模型，包括 CE-T1 放射组学特征、T2-WI 放射组学特征、CE-T1+T2-WI 放射组学特征和 CE-T1 放射组学列线图。CE-T1+T2-WI 放射组学特征在区分 NPC 患者对 IC 的反应和非反应方面表现良好，在主要队列中获得了 0.940（95%CI，0.885-0.974）的 AUC、83.1%的敏感性、91.8%的特异性和 87.1%的准确性，在验证队列中获得了 0.952（95%CI，0.855-0.992）的 AUC、74.2%的敏感性、95.2%的特异性和 82.7%的准确性。
结论
基于 MRI 的放射组学模型可能有助于对接受 IC 的 NPC 患者进行个体化风险分层和治疗。
Zotero 插件