Ng Matthew Song Peng, Loke Ryan Wai Keong, Tan Melvin Kian Loong, Ng Yau Hong, Liau Zi Qiang Glen
Department of Orthopaedics, National University Hospital, National University Health System, Singapore, Singapore.
Department of Orthopaedics, Alexandra Hospital, National University Health System, Singapore, Singapore.
Arthroplasty. 2025 Aug 4;7(1):39. doi: 10.1186/s42836-025-00322-1.
Robotic Total Knee Arthroplasty (rTKA) has become increasingly popular. Intraoperative manual planning of femur and tibia implant positions in all degrees of freedom to achieve surgeon-defined targets and limits of bone cuts, gaps, and alignment is challenging. The final manually defined solution may not be optimal, and surgical duration increases significantly. We aim to demonstrate the effectiveness of our novel algorithm in terms of accuracy and surgical duration.
We developed a novel AI computational algorithm to optimize rTKA implant positioning in three-dimensional space. The initial parameters of 3D implant positioning and surgeon-defined target gaps and bone cuts are set. The algorithm determines permutations achieving ideal 3D implant positioning with ± 0.5 mm accuracy, ranking them by surgeon preference and evidence-based criteria. We compared accuracy in achieving surgeon-defined target gaps, intraoperative soft tissue balancing duration, and total surgical time.
A prospective study of 67 consecutive rTKA patients at a tertiary institution (Nov 2021-Dec 2023) was conducted. 25 patients (mean age 70.4 ± 7.34 years) had our algorithm used intraoperatively, while 42 (mean age 70.5 ± 6.90 years) did not. 92% of rTKAs using our algorithm achieved target gaps ± 1.5 mm, vs. 52% of non-algorithm rTKAs (P = 0.003). The average difference between surgeon-defined target gaps and final achieved gaps was 1.1 ± 0.5 mm in the algorithm group vs. 1.8 ± 1.0 mm in the non-algorithm group (P = 0.003). Soft tissue balancing duration was significantly shorter: 1.16 min ± 0.11 with algorithm use vs. 14.5 min ± 8.3 (P < 0.0001). Total surgical duration was also significantly lower: 38.4 min ± 14.9 vs. 73.7 min ± 19.6 (P = 0.0002).
Our novel AI algorithm significantly improves accuracy in achieving surgeon-defined target extension and flexion gaps while reducing soft tissue balancing and total surgical duration. This is highly promising for achieving both reproducibility and efficiency in rTKAs. Video Abstract.
机器人全膝关节置换术（rTKA）越来越受欢迎。在所有自由度上对股骨和胫骨植入物位置进行术中手动规划，以实现外科医生定义的截骨、间隙和对线目标及限制具有挑战性。最终手动确定的方案可能不是最优的，并且手术时间会显著增加。我们旨在证明我们的新算法在准确性和手术时间方面的有效性。
我们开发了一种新颖的人工智能计算算法，以优化三维空间中的rTKA植入物定位。设置3D植入物定位的初始参数以及外科医生定义的目标间隙和截骨。该算法确定实现理想3D植入物定位（精度为±0.5毫米）的排列，并根据外科医生的偏好和循证标准对其进行排序。我们比较了实现外科医生定义的目标间隙的准确性、术中软组织平衡时间和总手术时间。
在一家三级医疗机构（2021年11月至2023年12月）对67例连续的rTKA患者进行了前瞻性研究。25例患者（平均年龄70.4±7.34岁）术中使用了我们的算法，而42例患者（平均年龄70.5±6.90岁）未使用。使用我们算法的rTKA中有92%实现了±1.5毫米的目标间隙，而非算法rTKA为52%（P = 0.003）。算法组中外科医生定义的目标间隙与最终实现的间隙之间的平均差异为1.1±0.5毫米，非算法组为1.8±1.0毫米（P = 0.003）。软组织平衡时间显著缩短：使用算法时为1.16分钟±0.11，而未使用算法时为14.5分钟±8.3（P < 0.0001）。总手术时间也显著降低：38.4分钟±14.9，而未使用算法时为73.7分钟±19.6（P = 0.0002）。
我们的新型人工智能算法在实现外科医生定义的目标伸直和屈曲间隙方面显著提高了准确性，同时减少了软组织平衡和总手术时间。这对于在rTKA中实现可重复性和效率非常有前景。视频摘要。
