摘要
目的探讨人工智能(Artificial Intelligence,AI)软件识别定位Ruedi-AugowerⅡ型Pilon骨折的精准性。方法回顾性纳入2020年1月至2024年8月Ruedi-AugowerⅡ型Pilon骨折DICOM格式CT数据20例,随机编号后,分别由1名骨科高年资医师和低年资医师在院内PACS系统和AI软件中识读分析,两者识别出需要复位的骨折块并经判定一致后,在不同系统中测量出每一骨折块在X轴(内外)、Y轴(上下)、Z轴(前后)上的位移、旋转角度,同时用AI软件得出骨折分型和骨折点信息。结果软件识别骨折分型准确率为70%;2组测量骨折块X、Y、Z轴位移、旋转角度等比较差异均无统计学意义(P>0.05),且2种测量方法的一致性较好。结论AI软件测量Ruedi-AugowerⅡ型Pilon骨折位移信息效能与人工一致,验证了其识别定位的精准性;其模拟复位测量的方式具有开创性,可为术前规划复位方案提供指导。
Objective To evaluate the accuracy of artificial intelligence(AI)software in identifying and localizing Ruedi-Augower typeⅡPilon fractures.Methods A retrospective analysis was conducted on 20 cases of Ruedi-Augower typeⅡPilon fractures with DICOM format CT data collected from January 2020 to August 2024.After random numbering,the data were analyzed by a senior orthopedic surgeon and a junior orthopedic resident using both the hospital’s PACS system and AI software.Once consensus was reached on the fracture fragments requiring reduction,measurements of displacement and rotation angles along the X-axis(medial-lateral),Y-axis(superior-inferior),and Z-axis(anterior-posterior)were performed in both systems.Additionally,the AI software was used to provide information on fracture classification and fracture point identification.Results The software achieved an accuracy rate of 70%for fracture classification.No statistically significant differences were observed between the two groups in terms of X,Y,and Z-axis displacements and rotation angles of the fracture fragments(P>0.05).The consistency between the two measurement methods was excellent.Conclusion The performance of the AI software in measuring displacement information for Ruedi-Augower typeⅡPilon fractures is comparable to manual measurement,validating its precision in identification and localization.Its innovative approach to simulating reduction measurements can provide valuable guidance for preoperative planning of reduction strategies.
作者
尹晓冬
成永忠
陈洋
闫威
孙海滨
李克
曾凡高
YIN Xiaodong;CHENG Yongzhong;CHEN Yang;YAN Wei;SUN Haibin;LI Ke;ZENG Fangao(Department of Trauma Surgery One,Wangjing Hospital of the China Academy of Chinese Medical Sciences,Beijing 100102,China;Department of Radiology,Wangjing Hospital of the China Academy of Chinese Medical Sciences,Beijing 100102,China;Department of Trauma Orthopedics,Dushan Branch of Nanyang Traditional Chinese Medicine Hospital,Nanyang Henan 473000,China;Department of Foot and Ankle Surgery,Dushan Branch of Nanyang Traditional Chinese Medicine Hospital,Nanyang Henan 473000,China)
出处
《中国医疗设备》
2025年第2期130-136,共7页
China Medical Devices
基金
中国中医科学院望京医院高水平中医医院建设项目中医药临床循证研究专项(WJYY-XZKT-2023-02)
国家自然科学基金面上项目(82274561)。
关键词
人工智能
骨折分型
胫骨骨折
计算机断层扫描(CT)
软件识别
骨折分型
DICOM格式
artificial intelligence
fracture classification
tibial fractures
computed tomography(CT)
software identification
fracture classification
DICOM format
