目的 探索基于深度学习的计算机辅助诊断系统(DL-CAD)在协助放射科医师进行肋骨骨折诊断方面的应用价值。方法 选择2022年1- 12月首都医科大学大兴教学医院因急性胸部外伤行胸部CT检查的患者330例共7920根肋骨,由低年资和高年资医师分别进行两轮(传统人工模式和DL-CAD辅助模式)阅片,并对诊断的敏感度、特异度、曲线下面积(AUC)值、阅片时间以及诊断信心等方面的差异进行统计分析。结果 共确认975处骨折,其中450处非错位骨折,525处错位骨折。两种阅片模式的敏感度、阅片时间、诊断信心及AUC值存在统计学差异(P<0.05),而特异度无统计学差异(P>0.05)。低年资及高年资医师两种模式阅片诊断错位性骨折及非错位性骨折的敏感度、AUC均存在统计学差异(P <0.05),特异度无统计学差异(P>0.05),其中诊断非错位性骨折的敏感度及AUC值显著提升(39.33% vs. 76.22%, 41.67% vs. 77.33%; 0.658 vs. 0.844, 0.671 vs. 0.861)。结论 DL_CAD能够协助放射科医师提高诊断肋骨骨折的效能,缩短阅片时间,增强诊断信心,有利于临床工作的开展。DL_CAD辅助后低年资和高年资医师的诊断效能趋同。
Abstract
Objective To explore the application value of deep learning-based computer-aided diagnosis system(DL-CAD) in assisting radiologists to diagnose rib fractures.Methods A total of 330 patients with 7920 ribs who underwent chest CT examination due to acute chest trauma in Daxing Teaching Hospital of Capital Medical University from January 2022 to December 2022 were selected. Two rounds of image reading (traditional manual mode and DL-CAD assisted mode) were performed by junior and senior radiologists respectively. The differences in sensitivity, specificity, AUC value, reading time, and diagnostic confidence were statistically analyzed.Results A total of 975 fractures were confirmed, including 450 non-displaced fractures and 525 displaced fractures. There were statistically significant differences in sensitivity, reading time, diagnostic confidence, and AUC value between the two modes (P<0.05), but no statistically significant difference in specificity (P>0.05). There were statistical differences in sensitivity and AUC in the diagnosis of displaced fractures and non-displaced fractures between the two modes of reading (P<0.05), and no statistically different in specificity (P>0.05). The sensitivity and AUC value in the diagnosis of non-displaced fractures were significantly improved in DL-CAD assisted mode (39.33% vs. 76.22%; 0.658 vs. 0.844; 41.67% vs. 77.33%; 0.671 vs. 0.61).Conclusions DL-CAD can help radiologists improve the efficiency of the diagnosis of rib fractures, shorten the reading time, and enhance the diagnostic confidence, which is conducive to the development of clinical work. DL_CAD-assisted diagnosis results in similar diagnostic performance between junior and senior radiologists.
关键词
肋骨骨折 /
深度学习 /
计算机体层成像 /
计算机辅助诊断
Key words
rib fracture /
deep learning /
computed tomography /
computer-aided diagnosis
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 谭 辉,田占雨,潘 宁,等. 基于深度学习的计算机辅助诊断系统在提高急性肋骨骨折诊断效能上的价值[J]. 临床放射学杂志,2020,39(12):2493-2497.
[2] 朱雅茹,祁 良,徐 磊,等. 人工智能软件辅助诊断新鲜肋骨骨折的效能评估[J]. 南京医科大学学报(自然科学版),2022,42(2):227-232.
[3] 史丹迪,张忠伟,胡铁波,等. 人工智能辅助放射科住院医师诊断肋骨新鲜骨折的应用价值[J]. 现代实用医学,2022,34(10):1308-1310.
[4] Choi J, Alawa J, Tennakoon L, et al. DeepBackRib: Deep learning to understand factors associated with readmissions after rib fractures[J]. J Trauma Acute Care Surg,2022,93(6):757-761.
[5] Niiya A, Murakami K, Kobayashi R, et al. Development of an artificial intelligence-assisted computed tomography diagnosis technology for rib fracture and evaluation of its clinical usefulness[J]. Sci Rep,2022,12(1):8363.
[6] Weikert T, Noordtzij L A, Bremerich J, et al. Assessment of a deep learning algorithm for the detection of rib fractures on whole-body trauma computed tomography[J]. Korean J Radiol,2020,21(7):891-899.
[7] 张武平,王春风,王 琳,等. 人工智能辅助在提高CT肋骨骨折检出率作用分析[J]. 影像研究与医学应用,2020,4(13):211-212.
[8] 熊 山,陈 博,毛 杰,等. 基于深度学习的计算机辅助诊断系统在肋骨骨折诊断中的应用[J]. CT理论与应用研究,2022,31(5):617-622.
[9] 张 倩,王大为,隋 时,等. 深度学习计算机辅助诊断系统检测胸部DR肋骨骨折的验证性研究[J]. 中国临床医学影像杂志,2022,33(6):430-434.
[10] 曹甲甲,张 阳,卞士中,等. 肋骨骨折法医临床学鉴定的初步认识[J]. 法医学杂志,2018,34(2):181-185.
[11] Yang C, Wang J, Xu J, et al. Development and assessment of deep learning system for the location and classification of rib fractures via computed tomography[J]. Eur J Radiol,2022,154:110434.
[12] Yao L, Guan X, Song X, et al. Rib fracture detection system based on deep learning[J]. Sci Rep,2021,11(1):23513.
[13] Jin L, Yang J, Kuang K, et al. Deep-learning-assisted detection and segmentation of rib fractures from CT scans: development and validation of FracNet[J]. EBioMed,2020,62:103106.
[14] Zhou Q Q, Tang W, Wang J, et al. Automatic detection and classification of rib fractures based on patients' CT images and clinical information via convolutional neural network[J]. Eur Radiol,2021,31(6):3815-3825.
PDF(3226 KB)
