基于国产人工智能软件对肋骨骨折诊断效能应用研究
Research on the Application of Artificial Intelligence Software in Rib Fracture Diagnosis
摘要: 目的:评估人工智能辅助分析软件(uAI Portal)对肋骨骨折的诊断效能。方法:回顾性收集2020年9月~10月在我院就诊的肋骨骨折患者102例。所有患者均经uAI Portal软件及放射科医师对肋骨骨折进行评估,肋骨骨折的金标准由两名高年资放射科医师共同决定。评估AI诊断肋骨骨折与金标准的一致性,计算AI与人工诊断的准确度、灵敏度、特异度、误诊率及漏诊率,并比较两种方法对肋骨骨折的诊断效能。结果:uAI Porta诊断结果与金标准之间差异无统计学意义(P > 0.05),两者之间呈高度一致性(KAPPA指数:0.965,P < 0.05),uAI Portal对肋骨骨折诊断的准确性、灵敏度、特异度、漏诊率及误诊率分别为98.9%、95.7%、99.7%、0.3%、4.2%。uAI Portal在1~3、4~6、7~9肋段分布的肋骨骨折与放射科医师的诊断效能差异具有统计学意义(P < 0.05),在10~12肋段分布的肋骨骨折两者诊断效能差异无统计学意义(P > 0.05)。结论:uAI Portal软件对肋骨骨折的诊断具有较高的准确度、灵敏度和特异度以及较低的漏诊率和误诊率。同时,在1~9肋段分布的肋骨骨折uAI Portal诊断效能高于放射科医师,表明应用uAI Portal可以显著提高放射科医师对肋骨骨折的诊断效能,值得临床进一步推广应用。
Abstract: Objective: To evaluate the effectiveness of artificial intelligent-assisted analysis software (UAI Portal) in the diagnosis of rib fractures. Methods: 102 patients with rib fractures in our hospital from September to October 2020 were collected retrospectively. All patients were evaluated for rib fractures by uAI Portal software and radiologists. The accuracy, sensitivity, specificity, misdiagnosis rate and missed diagnosis rate of the two methods were calculated, and the diagnostic efficacy of two methods for rib fracture was compared. Results: There was no significant difference between the results of uAI Portal and the gold standard, and there was a high consistency between them (KAPPA index: 0.965, P < 0.05). The accuracy, sensitivity, specificity, missed diagnosis rate and misdiagnosis rate of uAI Portal in the diagnosis of rib fractures were 98.9%, 95.7%, 99.7%, 0.3% and 4.2%. There was significant difference in diagnostic efficiency between uAI Portal and radiologists in rib fractures of 1~3, 3~6 and 6~9 ribs (P < 0.05), but there was no significant difference between the two in rib fractures of 10~12 ribs (P > 0.05). Conclusion: uAI Portal software has high accuracy, sensitivity and specificity in the diagnosis of rib fracture, as well as low missed diagnosis rate and misdiagnosis rate. At the same time, the diagnostic efficiency of uAI Portal in rib fractures distributed in 1~9 ribs was significantly higher than that of radiologists. It shows that the application of uAI-Bone Care can significantly improve the diagnostic efficiency of radiologists in the diagnosis of rib fractures, which is worthy of further clinical application.
文章引用:王雄慧, 雷禹, 牛媛, 李静, 朱彦瑾, 姬星, 黄晓旗. 基于国产人工智能软件对肋骨骨折诊断效能应用研究[J]. 临床医学进展, 2022, 12(1): 404-411. https://doi.org/10.12677/ACM.2022.121061

参考文献

[1] Lin, F.C., Li, R.Y., Tung, Y.W., et al. (2016) A Morbidity, Mortality, Associated Injuries, and Management of Traumatic Rib Fractures. Journal of the Chinese Medical Association, 79, 329-334. [Google Scholar] [CrossRef] [PubMed]
[2] Henry, T.S., Donnelly, E.F., Boiselle, P.M., et al. (2019) ACR Appropriateness Criteria® Rib Fractures. Journal of the American College of Radiology, 16, S227-S234. [Google Scholar] [CrossRef] [PubMed]
[3] Maduka, A.K.I., Lin, G.J., Lim, W.W., et al. (2019) A Comprehensive Analysis of Traumatic Rib Fractures in an Acute General Hospital in Singapore. Journal of Emergencies, Trauma, and Shock, 12, 145-149. [Google Scholar] [CrossRef
[4] 茅怡铭, 吴长江. 老年多发肋骨骨折患者肺部并发症风险预测模型的建立[J]. 上海交通大学学报(医学版), 2019, 39(1): 69-72. [Google Scholar] [CrossRef
[5] Cho, S.H., Sung, Y.M. and Kim, M.S. (2012) Missed Rib Fractures on Evaluation of Initial Chest CT for Trauma Patients: Pattern Analysis and Diagnostic Value of Coronal Multiplanar Reconstruction Images with Multidetector Row CT. British Journal of Radiology, 85, e845-e850. [Google Scholar] [CrossRef] [PubMed]
[6] Van Laarrhoven, J., Hietbrink, F., Ferree, S., et al. (2019) Associated Thoracic Injury in Patients with a Clavicle Fracture: A Retrospective Analysis of 1461 Polytrauma Patients. European Journal of Trauma and Emergency Surgery, 45, 59-63. [Google Scholar] [CrossRef] [PubMed]
[7] 陆芳, 葛虓俊, 李骋, 等. CT肋骨拉伸后处理技术诊断肋骨骨折的价值[J]. 中华放射学杂志, 2017, 51(9): 695-698. [Google Scholar] [CrossRef
[8] 汪琼, 孙婷婷, 范鸿禹, 等. AI辅助软件对于提高不同年资规培医师肺结节检测能力的影响[J]. 国际医学放射学杂志, 2020, 43(4): 403-408. [Google Scholar] [CrossRef
[9] 卢暄, 卢平明, 张雪宁, 等. 64排螺旋CT后处理技术诊断肋骨骨折的应用价值[J]. 实用放射学杂志, 2015(4): 642-645. [Google Scholar] [CrossRef
[10] 周清清, 王佳硕, 唐雯, 等. 基于卷积神经网络成人肋骨骨折CT自动检测和分类的应用研究[J]. 影像诊断与介入放射学, 2020, 29(1): 27-31. [Google Scholar] [CrossRef
[11] 齐敏, 董志坚, 郑颖, 等. CT Bone Reading软件对肋骨骨折诊断效能的应用价值[J]. 实用放射学杂志, 2019, 35(9): 1476-1479. [Google Scholar] [CrossRef
[12] Ringl, H., Lazar, M., Töpker, M., et al. (2015) The Ribs Unfolded—A CT Visualization Algorithm for Fast Detection of Rib Fractures: Effect on Sensitivity and Specificity in Trauma Patients. European Radiology, 25, 1865-1874. [Google Scholar] [CrossRef] [PubMed]
[13] 郭琰. 肋骨曲面重建方法探究[J]. 实用放射学杂志, 2018, 34(8): 1302-1303. [Google Scholar] [CrossRef