Abstract:

Objective　To explore the clinical application value of Artificial Intelligence (AI) technology in preoperative planning of unicompartmental knee arthroplasty. Methods　This study is a retrospective controlled analysis. A total of 35 patients who underwent unicompartmental knee arthroplasty at Huangshi Central Hospital from July 2021 to July 2023 were selected and divided into an experimental group and a control group based on the preoperative planning method used. Among the 14 patients in the experimental group, 6 were male and 8 were female, the age was (62.43±3.55) years old, AI technology was used for preoperative planning. In the control group, there were 7 males and 14 females in 21 patients, the age was (62.10±2.86) years old, Preoperative planning was performed according to the traditional X-ray method. The surgical time, intraoperative blood loss, postoperative drainage volume, prosthesis conformity rate (the preoperatively predicted sizes and models of the prosthesis and spacer matching exactly with those used intraoperatively), as well as the American Hospital Specialty (HHS) knee scores, Visual Analog Scale (VAS) scores, and improvement in knee range of motion at 3 months and 1 year postoperatively were recorded and compared. The χ² test and t test were used for statistical analysis. Results　The surgical time, intraoperative blood loss, and postoperative drainage volume in the experimental group were (59.07±3.36) minutes, (107.29±10.16) ml, and (78.93±10.95) ml, The control group were (69.57±2.11) minutes, (123.38±7.65) ml, and (95.19±12.49) ml, with all differences being statistically significant (t=11.378, 5.347,and 3.958, all P<0.001). At 3 months and 1 year postoperatively, the HHS knee scores and the range of motion of the knee was greater in the experimental group[(79.93±1.14) points vs. (76.81±1.54) points,(119.57±4.62)° vs. (111.76±6.25)°,(92.36±1.87) points vs. (90.48±1.25) points,(130.29±4.60)° vs. (120.57±4.98)°], and VAS scores in the experimental group were lower than those in the control group [(1.71±0.47) points vs. (2.43±0.68) points, (1.14±0.36) points vs. (1.86±0.48) points], with all differences being statistically significant (all P<0.05). The conformity rates for the femoral side, tibial side, and spacer in the experimental group were higher than those in the control group [92.86% (13/14) vs. 42.86% (9/21), 92.86% (13/14) vs. 38.10% (8/21), 85.71% (12/14) vs. 33.33% (7/21), all P<0.05]. Conclusion　The application of AI technology in preoperative planning for unicompartmental knee arthroplasty can accurately predict the model of the femoral and tibial prostheses as well as the size of the spacer, improve prosthesis conformity rates, reduce intraoperative blood loss and postoperative drainage volume, alleviate postoperative knee pain symptoms, and enhance knee range of motion, demonstrating significant clinical guidance value.

Key words: Unicompartmental knee arthroplasty,  , Knee osteoarthritis,  , Preoperative planning,  , Visualization,  , Artificial intelligence

摘要：

目的　探索人工智能（AI）技术在膝单髁置换术前规划中的临床应用价值。方法　本研究为回顾性对照分析。选取2021年7月至2023年7月在黄石市中心医院行膝单髁置换术的35例患者，根据术前规划的方法不同分为试验组和对照组。试验组14例患者中男6例，女8例，年龄（62.43±3.55）岁，应用AI技术进行术前规划。对照组21例患者中男7例，女14例，年龄（62.10±2.86）岁，按照传统X线片方法进行术前规划。记录并比较两组患者手术时间、术中出血量、术后引流量、假体符合率（术前预测假体、垫片的大小型号与术中使用完全一致）以及术后3个月、1年美国特种外科医院（HHS）膝关节评分、视觉模拟评分法（VAS）评分、膝关节活动度改善情况。采用χ2检验、t检验进行对比分析。结果　试验组手术时间、术中出血量、术后引流量分别为（59.07±3.36）min、（107.29±10.16）ml、（78.93±10.95）ml，对照组分别为（69.57±2.11）min、（123.38±7.65）ml、（95.19±12.49）ml，差异均有统计学意义（t=11.378、5.347、3.958，均P<0.001）。术后3个月、1年，试验组的HHS膝关节评分和膝关节活动度均高对照组［（79.93±1.14）分比（76.81±1.54）分和（119.57±4.62）°比（111.76±6.25）°、（92.36±1.87）分比（90.48±1.25）分和（130.29±4.60）°比（120.57±4.98）°］，VAS评分均低于对照组［（1.71±0.47）分比（2.43±0.68）分、（1.14±0.36）分比（1.86±0.48）分］，差异均有统计学意义（均P<0.05）。试验组的股骨侧、胫骨侧、垫片符合率均高于对照组［92.86%（13/14）比42.86%（9/21）、92.86%（13/14）比38.10%（8/21）、85.71%（12/14）比33.33%（7/21），均P<0.05］。结论　应用AI技术进行膝单髁置换术前规划，可精准预测股骨侧、胫骨侧假体型号及垫片大小，提高假体符合率，减少术中出血及术后引流量，改善患者术后膝关节疼痛症状，提高膝关节活动度，具有临床指导意义。

关键词: 膝单髁置换术,  ,  , 膝骨关节炎,  ,  , 术前规划,  ,  , 可视化,  ,  , 人工智能