住培医师应用人工智能提升胸部CT肋骨骨折检出率的价值

The value of applying artificial intelligence for resident training physicians to improve the effectiveness of rib fracture detection on chest CT

目的:探讨住培医师应用人工智能(artificial intelligence,AI)提升胸部CT肋骨骨折检出效能的价值。方法:回顾性分析2021年1月—11月在梅州市人民医院行胸部CT检查的57例肋骨骨折患者资料。住培医师首先采用单独阅片进行图像分析,间隔两周洗脱期后应用人工智能阅片。比较住培医师单独阅片和应用AI阅片对肋骨骨折检出率的差异性;比较住培医师单独阅片和应用AI阅片的诊断时间的差异性。结果:住培医师A应用AI阅片对肋骨骨折的检出率为92.00%,高于单独阅片的80.00%(P<0.05);住培医师B应用AI阅片对肋骨骨折的检出率为94.40%,高于单独阅片的83.20%(P<0.05)。住培医师A和B在应用AI阅片后检出肋骨骨折的诊断时间均短于单独阅片(P<0.05)。结论:住培医师应用人工智能有助于提升胸部CT肋骨骨折检出率,同时能够显著缩短诊断时间。

Objective To explore the value of applying artificial intelligence(AI)by resident physicians to enhance the efficacy of rib fracture detection on chest CT.Methods The data of 57 patients with rib fractures who underwent chest CT examination in Meizhou People's Hospital from January to November 2021 were retrospectively analyzed.The resident training physicians first used independent reading for image analysis and applied artificial intelligence reading after a two-week elution interval.Comparison of the difference in the detection rate of rib fractures between the independent and AI reading by the resident trainers;Comparison of diagnostic time differences between the independent and AI reading by the resident trainers.Results The detection rate of rib fracture was 92.00%for resident training physician A after AI reading,which was significantly higher than that of 80.00%for independent reading(P<0.05).The detection rate of rib fractures by resident training physician B after applying AI reading was 94.40%,which was significantly higher than the detection rate of 83.20%by independent reading(P<0.05).The diagnostic time to detect rib fractures was significantly shorter for resident training physician A and B after applying AI to read images than for independent reading(P<0.05,respectively).Conclusion The application of AI by resident training physicians helps to improve the detection rate of rib fractures on chest CT,and at the same time can significantly shorten the diagnosis time.