Scholarlink

Background <p>Accurate implant sizing is critical for clinical outcomes in Oxford unicompartmental knee arthroplasty (UKA), yet reliable preoperative planning remains challenging. This study aimed to evaluate whether AI-assisted preoperative planning improves component sizing accuracy compared with conventional intraoperative instrumentation and to explore factors associated with component coverage.</p> Methods <p>A single-center retrospective cohort study was conducted including 190 patients (190 knees) undergoing primary Oxford UKA between February 2022 and January 2024. Patients were categorized into the AI-assisted group (<i>n</i> = 94) or the conventional group (<i>n</i> = 96). Radiographic sizing accuracy of femoral and tibial components was assessed postoperatively. Clinical outcomes were evaluated using the HSS score at predefined intervals of 6, 12, and 24 months postoperatively. Logistic regression analysis was performed to identify factors associated with inaccurate sizing in the AI-assisted group.</p> Results <p>The AI-assisted group demonstrated significantly higher radiographic sizing accuracy than the conventional group for the femoral component (81.91 vs. 63.54%, <i>P</i> = 0.005), tibial component (85.12 vs. 67.71%, <i>P</i> = 0.005), and both components (70.21 vs. 51.04%, <i>P</i> = 0.007). Patients with accurate sizing of both components achieved higher HSS scores at 24 months compared with those with at least one inaccurately sized component (92.50 ± 4.00 vs. 89.32 ± 3.96, <i>P</i> &lt; 0.001). Within the AI-assisted group, agreement between planned and actual posterior condylar resection (OR = 0.249, <i>P</i> = 0.025) and acceptable femoral sagittal angle (OR = 0.252, <i>P</i> = 0.020) were independent predictors of femoral sizing accuracy. No significant independent predictors were identified for tibial inaccurate sizing.</p> Conclusions <p>AI-assisted preoperative planning was associated with improved radiographic component sizing accuracy in Oxford UKA compared with conventional instrumentation. While no significant differences in HSS scores were observed at 6 and 12 months, patients with accurate sizing of both components demonstrated significantly better functional recovery at the 24-month follow-up. The reliability of AI-assisted femoral sizing depends on precise intraoperative execution of planned resection and alignment parameters.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AI-assisted preoperative planning improves component sizing accuracy in Oxford unicompartmental knee arthroplasty: a retrospective cohort study with 24-month follow-up

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月30日

查看全文
  • Fengrui Liu,
  • Tong Li,
  • Mengchen Yuan,
  • Jiaxin Sun,
  • Guoyuan Sun,
  • Yaxing Duan,
  • Zhengyuan Bai,
  • Jinrui Sun,
  • Xingyu Liu,
  • Yiling Zhang,
  • Qidong Zhang

摘要

Background

Accurate implant sizing is critical for clinical outcomes in Oxford unicompartmental knee arthroplasty (UKA), yet reliable preoperative planning remains challenging. This study aimed to evaluate whether AI-assisted preoperative planning improves component sizing accuracy compared with conventional intraoperative instrumentation and to explore factors associated with component coverage.

Methods

A single-center retrospective cohort study was conducted including 190 patients (190 knees) undergoing primary Oxford UKA between February 2022 and January 2024. Patients were categorized into the AI-assisted group (n = 94) or the conventional group (n = 96). Radiographic sizing accuracy of femoral and tibial components was assessed postoperatively. Clinical outcomes were evaluated using the HSS score at predefined intervals of 6, 12, and 24 months postoperatively. Logistic regression analysis was performed to identify factors associated with inaccurate sizing in the AI-assisted group.

Results

The AI-assisted group demonstrated significantly higher radiographic sizing accuracy than the conventional group for the femoral component (81.91 vs. 63.54%, P = 0.005), tibial component (85.12 vs. 67.71%, P = 0.005), and both components (70.21 vs. 51.04%, P = 0.007). Patients with accurate sizing of both components achieved higher HSS scores at 24 months compared with those with at least one inaccurately sized component (92.50 ± 4.00 vs. 89.32 ± 3.96, P < 0.001). Within the AI-assisted group, agreement between planned and actual posterior condylar resection (OR = 0.249, P = 0.025) and acceptable femoral sagittal angle (OR = 0.252, P = 0.020) were independent predictors of femoral sizing accuracy. No significant independent predictors were identified for tibial inaccurate sizing.

Conclusions

AI-assisted preoperative planning was associated with improved radiographic component sizing accuracy in Oxford UKA compared with conventional instrumentation. While no significant differences in HSS scores were observed at 6 and 12 months, patients with accurate sizing of both components demonstrated significantly better functional recovery at the 24-month follow-up. The reliability of AI-assisted femoral sizing depends on precise intraoperative execution of planned resection and alignment parameters.