分析人工智能技术在多层螺旋CT中诊断肋骨骨折的效能价值

To Analyze the Value of Artificial Intelligence in Diagnosing Rib Fracture with Multi-Slice Spiral CT

目的:探讨人工智能(AI)肋骨骨折检测系统在多层螺旋CT中提高诊断肋骨骨折效能方面的作用。方法:选取2021年5月~2022年5月沈阳市骨科医院经64排螺旋CT扫描的50例肋骨骨折阳性和50例肋骨骨折阴性的患者作为观察对象,然后分别运用AI影像分析系统(AI组)、人工组及人工+AI组对比分析肋骨的特征,得出肋骨骨折的数量,然后统计三组诊断肋骨骨折的检出时间,假阳性率,敏感性和准确率。结果:AI检测出肋骨骨折的平均时间为(63±18)s,假阳性率4.56%,敏感性87.75%,准确率95.13%;人工检测出肋骨骨折的平均时间为(203±68)s,假阳性率1.30%,敏感性84.69%,准确率97.91%;AI辅助人工检测出肋骨骨折的平均时间为(136±28)s,假阳性率0.96%,敏感性92.85%,准确率98.79%。结论:AI技术在肋骨骨折中具有较高的敏感性和诊断准确率,虽然存在一定的假阳性率,但可以辅助人工,减少阅片时间,在提高诊断质量和工作效率上效果显著。

Objective:To evaluate the effectiveness of artificial intelligence(AI)rib fracture detection system in diagnosing rib fracture with multi-slice spiral CT(MSCT).Methods:From May 2021 to May 2022,50 patients with rib fracture and 50 patients with rib fracture were by 64-slice spiral CT scan,then,using AI image analysis system(AI group),artificial group and artificial+AI group to compare and analyze the characteristics of rib,get the number of rib fracture,and then count the detection time of rib fracture in three groups,false positive rate,sensitivity and accuracy.Results:The average time for AI to detect rib fracture was(63±18)s,the false positive rate was 4.56%,the sensitivity was 87.75%,the accuracy was 95.13%,the average time for AI to detect rib fracture was(203±68)s,the false positive rate was 1.30%,the sensitivity was 84.69%,the average time of artificial rib fracture detection by AI was(136±28)s,the false positive rate was 0.96%,the sensitivity was 92.85%,the accuracy was 98.79%.Conclusion:AI technique has a high sensitivity and accuracy in diagnosis of RIB fracture.Although there is a certain false positive rate,it can be used as an assistant in manual operation and reduce the time of reading films,it has remarkable effect in improving the quality of diagnosis and working efficiency.