Page 83 - 南京医科大学自然版
P. 83
第46卷第9期 万丽娟,李金凯,苏国义,等. 基于临床及剂量学特征构建模型预测鼻咽癌放疗后颞叶损伤的研究[J].
2026年9月 南京医科大学学报(自然科学版),2026,46(9):1348-1355 ·1353 ·
Radiotherapy plan and follow⁃up images of a patient with nasopharyngeal carcinoma staged as T2N3M1(stage IVB). The dosevolume histogram of
the left temporal lobe shows D 1 cm 3 =5 642.3 cGy and V70Gy=0%. Followup MRI images obtained 22 months after radiotherapy showed no radiationinduced
injury in the temporal lobe.
图2 1例未发生TLI的鼻咽癌患者的放疗计划和MRI影像
Figure 2 Radiotherapy plan and MRI images of a nasopharyngeal carcinoma patient without TLI
A B C
0.15
1.0 Clinical model
1.0
Combined model
0.8 0.8 0.10 All
None
Sensitivity 0.6 Observed probability 0.6 Net benefit 0.05
0.4
0.4
Clinical model
(AUC=0.635) Ideal 0
0.2 0.2
Combined model Apparent
(AUC=0.853) 0 Bias⁃corrected -0.05
0
.0 0.2 0.4 0.6 0.8 1.0 .0 0.2 0.4 0.6 0.8 1.0 .0 0.1 0.2 0.3 0.4 0.5
1-Specificity Predicted probability High risk threshold
D 0 20 40 60 80 100
Points
2 4
T
1 3
D1cc
45 55 65 75 85
V70 Gy
0 2 4 6 8 10
Total points
0 40 80 120 160
Risk of TLI
0.01 0.05 0.20 0.80
A:ROC curves of the clinical model and combined model. B:Calibration curve of the combined model. C:Decision curve analysis results of the
clinical model and combined model. D:Nomogram of the combined model.
图3 365例鼻咽癌放疗患者的临床模型及联合模型对放射性TLI的预测结果
Figure 3 Prediction results of the clinical model and combined model for radiation⁃induced TLI in 365 patients with naso⁃
pharyngeal carcinoma undergoing radiotherapy
技术(包括IMRT和VMAT)的鼻咽癌患者,其颞叶的 数进行单因素和多因素分析发现 T分期、 D
1 cm 3 、V70 Gy
最佳剂量⁃容积限制尚未完全确定 [23] 。使用临床剂 是放射性TLI的独立预测因素。在临床特征统计分
量学模型和人工智能模型分析数据 [24] ,为放疗计划 析时发现高血压、糖尿病等基础疾病在 TLI 组与非
提出适当的限制,实现安全的剂量递增。 损伤组间无显著统计学差异,与既往部分研究结果
本研究对鼻咽癌患者的临床特征及剂量学参 相符。然而在一项基于定量 MRI 生物标志物和临

