计算机辅助诊断系统在甲状腺结节超声诊断中的应用

[摘 要] 目的 评估计算机辅助诊断（Computer-aided Diagnosis ,CAD）系统在甲状腺超声检查中的诊断效能及临床应用价值。方法 选取2018年8月至2019年1月在中南大学湘雅三医院进行甲状腺超声检查并行手术切除的171例患者，共205个甲状腺结节。分别采用CAD及4名不同经验水平的超声医师对205例甲状腺结节的超声图像进行分析并根据美国放射学会（ACR）的TI-RADS指南进行分类，然后对4名不同经验水平的超声医师结合CAD也进行了研究。以手术病理结果为金标准,评估CAD系统在鉴别甲状腺良恶性结节中的诊断效能，以及CAD对不同经验水平的超声医师的影响。结果 CAD系统的使用提高了4名超声医师对甲状腺结节鉴别诊断的敏感性及AUC值(结合CAD系统vs未结合CAD系统:灵敏度：超声医师A, 93.10% vs 87.93%;超声医师B, 90.52% vs 84.48%;超声医师C, 85.34% vs 78.45%;超声医师D, 75.00% vs 66.38%，AUC值：超声医师A, 0.95 vs 0.94;超声医师B, 0.93 vs 0.92;超声医师C, 0.86vs 0.81;超声医师D, 0.86 vs 0.70)，差异均有统计学意义（P均＜0.05）。然而CAD系统的低特异性(73.03%)仅对超声医师C、D有显著性改善(P均＜0.05)。CAD系统对甲状腺恶性肿瘤的诊断敏感性与具有5年经验的超声医师相似，差异无统计学意义（P=1.00），但CAD系统的特异性较低。结论 CAD鉴别诊断甲状腺结节的敏感性较高，但特异性较低。结合CAD可有效提高初级医师甲状腺结节的超声诊断水平，也能提高高年资医师对甲状腺结节鉴别诊断的敏感性。

Application of computer-aided diagnosis system in ultrasonic diagnosis of thyroid nodules

[Abstract] Objective To evaluate the diagnostic efficiency and clinical application value of Computer-aided Diagnosis (CAD) system in thyroid ultrasound examination. Methods From August 2018 to January 2019, 171 patients with 205 thyroid nodules who underwent thyroid ultrasonography and surgical resection in The Third Xiangya Hospital of Central South University. The ultrasound images of 205 cases of thyroid nodules were analyzed by CAD and 4 radiologists with different levels of experience, and classified according to the TI-RADS guidelines of the American College of Radiology(ACR),then 4 radiologists with different levels of experience combined with CAD were also studied. Based on the results of surgery and pathology as the gold standard, the diagnostic efficiency of CAD system and 4 radiologists with different experience levels in differentiating benign and malignant thyroid nodules was compared, and the influence of CAD on radiologists with different experience levels was evaluated. Results The use of CAD system improved the sensitivity and area under curve values of 4 radiologists in differential diagnosis of thyroid nodules (CAD vs without CAD: sensitivity:radiologist A,93.10%vs87.93%; radiologist B, 90.52%vs84.48%;radiologist C,85.34%vs78.45%;radiologist D,75.00%vs66.38%,AUROC:radiologist A,0.95vs0.94;radiologist B,0.93vs0.92;radiologist C,0.86vs0.81;radiologist D,0.86vs0.70), the difference was statistically significant (P < 0.05). However, the low specificity of CAD system (73.03%) was only significantly improved for radiologist C and D(P < 0.05). The diagnostic sensitivity of CAD system to thyroid malignant tumor is similar to that of radiologists with 5 years of experience, and the difference is not statistically significant (P =1.00), but the specificity of CAD system is low. Conclusion The sensitivity of CAD in differential diagnosis of thyroid nodules is high, but the specificity is low. The combination of CAD can effectively improve the ultrasonic diagnosis level of thyroid nodules in junior radiologists,and can also improve the sensitivity of senior radiologists to differential diagnosis of thyroid nodules.