Abstract 【Objective】 To explore the application value of deep learning-based artificial intelligence (AI) software in the diagnosis of CT of fresh rib fractures.【Methods】A total of 259 qualified fresh fracture patients admitted to our hospital from June 2022 to September 2023 were selected and their imaging data were introduced to the AI workstation; Based on a 8-year attending physician in the department of radiology and AI software to diagnose rib fractures, the group A was diagnosed separately by AI software, and the group B and the group AB were diagnosed by physician combined with AI software. The number, site, and classification of fracture lesions were recorded in the three groups; The results of the three groups were analyzed with the reference standard to compare the sensitivity, misdiagnosis rate and detection rate of incomplete fractures in the three groups; Mean time of diagnosis was compared between the groups B and the group AB. 【Results】There were 563 fractures in 259 patients; There were 385 complete fractures and 178 incomplete fractures. The sensitivity of fracture detection in the group A, the group B and the group AB was 91.47%, 83.13%, and 92.54%, respectively. The sensitivity comparison of the three groups showed statistically significant differences (P<0.05).The sensitivity of fractures in the group A and the group AB was higher than that in the group B, significant (P<0.05). The misdiagnosis rates in the group A, the group B and the group AB were 20.03%, 13.17% and 12.58%, respectively, which was statistically significant (P<0.05).The group A had the highest misdiagnosis rate, significantly higher than that of the group AB and the group B, with a statistically significant difference (all P<0.05). The detection rate of incomplete fractures in the group A, the group B and the group AB was 75.28%, 50.56% and 76.40%, respectively, and the difference was statistically significant (P<0.05).The mean diagnosis time in the group AB was shorter than that of the group B , with significant difference(P<0.05).【Conclusion】The joint diagnosis between doctors and AI improves the sensitivity of the detection of fresh rib fractures, which is mainly reflected in the detection rate of incomplete fractures, and can also significantly shorten the diagnosis time and improve the work efficiency.
[1] PEEK J,BEKS R B,HIETBRINK F,et al.Epidemiology and outcome of rib fractures:a nationwide study in the Netherlands[J].Eur J Trauma Emerg Surg,2022,48(1):265-271.
[2] 王瑶.多层螺旋CT多平面重建技术在创伤性肋骨骨折患者诊断中应用价值探析[J].实用医学影像杂志,2024,25(2):140.
[3] SAHINER B,CBAN H P,HADJIISKI L M,et al. Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans: analysis of an observer performance study by nodule size[J].Acad Radiol,2009,16(12):1518-1530.
[4] XIA Y,SHEN L,WEN L F,et al. Automated identification of dementia using FDG-PET imaging[J].Biomed Res Int,2014,2014:421743.
[5] 高歌,马帅,王霄英.计算机辅助诊断在医学影像诊断中的基本原理和应用进展[J].放射学实践,2016,31(12): 1127-1129.
[6] RINGL H, LAZAR M, TÖPKER M, et al. The ribs unfolded-a CT visualizantion algorithm for fast detection of rib fractures: effet on sensitivity and specifieity in trauma patients[J].Eur Radiol,2015,25(7):1865-1874.
[7] KASHIKURA Y, NAKAYAMA R, HIZUKURI A, et al. Improved differential diagnosis of breast masses on ultrasonographic imagey with a computer-aided diagnosis scheme for determining histological classifications[J].Acad Radiol,2013,20(4) :471-477.
[8] MATSUMOTO S,OHNO Y,AOKI T,et al. Computer-aided detection of lung nodules on multidetector CT in concurrent-reader and secondreader modes:a comparative study [J].Eur J Radiol,2013,82(8):1332-1337.
[9] 韩英,何生,姜增誉,等.人工智能在乳腺影像领域的应用现状[J].放射学实践,2019,34(7):813-816.
[10] 王成弟,郭际香,杨阳,等.利用深度学习技术辅助肺结节的人工智能检测[J].中国呼吸与危重监护杂志,2019,23(3):288- 294.
[11] 祝佳,关鑫江.隐匿性肋骨骨折52例多排 CT 扫描误诊漏诊分析[J].实用医药杂志,2018,35(1):36-38.
[12] 袁强,夏学文,魏文洲,等.胸部X线片与多层螺旋CT诊断肋骨骨折漏诊误诊分析[J].实用医学影像杂志,2017,18(3):185-187.
[13] 董浩,经齐峰,邱勇刚,等.基于深度学习人工智能辅助CT检测肋骨骨折的价值[J].浙江临床医学,2022,24(6):914-915.
2024, Vol. 41 
