Page 127 - 南京医科大学自然版

第45卷第8期 李 沁，丁志颖，宗晴晴，等. 基于超声构建的列线图模型在鉴别非平行位乳腺小肿块良恶性中的
2025年8月 应用价值［J］. 南京医科大学学报（自然科学版），2025，45（8）：1178-1185 ·1183 ·

0 10 20 30 40 50 60 70 80 90 100
Points

Age
20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Not circumscribed
Margin
Circumscribed
Intermediate
Elasticity
Soft Hard
4B
US⁃BI⁃RADS
4A
Total points
0 20 40 60 80 100 120 140 160 180 200 220
Risk of event
0.1 0.2 0.30.40.50.6 0.7 0.8 0.9
Nomogram to predict the malignancy in patients with small BI⁃RADS 4A，4B breast lesions that featured non⁃parallel orientation on ultrasound.
The nomogram was developed in the training set，which incorporated age，margin，elasticity and US⁃BI⁃RADS. The total points were calculated and pro⁃
jected at the bottom scale indicate the malignancy risk.
图1 基于训练集构建的列线图模型
Figure 1 Nomogram to predict the malignancy based on the training set

A B C
1.0 0.4 Nomogram
0.8 0.8 0.3 All
None
0.6
Sensitivity 0.6 AUC：0.846 Actrual probability 0.4 Apparrent Net benefit 0.2
0.1
0.4
0.2
Bias⁃corrected
Ideal 0
0.2 0
00 0.2 0.4 0.6 0.8 1.0
00 0.2 0.4 0.6 0.8 High risk threshold
0
Predicted Probability
00 0.2 0.4 0.6 0.8 1.0
B=400 Mean absolute 1∶100 1∶4 2∶3 3∶2 4∶1 100∶1
1-Specificity
repetitions，boot error=0.017（n=196） Cost：benefit ratio
D E F
1.0 0.4 Nomogram
0.8 0.6 0.3 All
None
Sensitivity 0.6 AUC：0.798 Actrual probability 0.8 Apparrent Net bbenefit 0.2
0.4
0.1
0.4
0.2
Ideal
0.2 Bias⁃corrected 0
0 00 0.2 0.4 0.6 0.8 1.0
0 0.2 0.4 0.6 0.8 High risk threshold
0
Predicted Probability
00 0.2 0.4 0.6 0.8 1.0
B=400 Mean absolute 1∶100 1∶4 2∶3 3∶2 4∶1 100∶1
1-Specificity
repetitions，boot error=0.038（n=73） Cost：benefit ratio
A：ROC curve of the nomogram in the training set. B：Calibration curves of the nomogram in the training set，with Hosmer⁃Lemeshow test showing
P=0.589. C：Decision curve analysis for the prediction model in the training set. D：ROC for nomogram in the validation set. E：Calibration curves for no⁃
mogram in validation set，with Hosmer⁃Lemeshow tests showing P=0.206. F：Decision curve analysis for the prediction model in the validation set.
图2 预测模型的鉴别效能及临床意义评估
Figure 2 Evaluation of the effectiveness of the predictive model
观察，因此导致本研究的选择偏倚，以及较高的总 综上，本研究回顾性分析了超声BI⁃RADS 4A及
体恶性率和较小的样本量。其次，本研究是单中心 4B类的非平行位生长的乳腺小肿块，基于临床及超
研究，BI⁃RADS分类也与本中心的医师经验相关，因 声特点构建了诊断预测模型。同时，进一步对乳腺
此增加外部队列的验证能提升模型的普适性及临 肿块进行了危险分层并提出诊疗建议，对于低危患
床实用性。 者，可采取密切的影像学随访，而对于高危患者，建