Abstract：
OBJECTIVE: Successful total hip arthroplasty relies on accurate preoperative planning. However, the conventional preoperative planning, a two-dimensional method using X-ray template, has shown poor reliability of predicting component size. To our knowledge, artificial intelligence technology assisted three-dimensional preoperative planning is promising to improve the accuracy of preoperative planning but there is a dearth of clinical evidence. Therefore, in this study we compared the prediction accuracy of these two maneuvers.
METHODS: We conducted a prospective study consisting of 117 consecutive patients who underwent a primary cementless total hip arthroplasty to compare the prediction accuracy of these two methods. The two-dimensional and artificial intelligence assisted three-dimensional planning results of the same patient were compared with the definitive implant size respectively.
RESULTS: The prediction accuracy of artificial intelligence assisted three-dimensional planning for cup and the stem sizes were 66.67% (78/117) and 65.81% (77/117), two-dimensional planning was 30.77% (36/117) and 37.61% (44/117) (p < 0.05). There were poor prediction results of two-dimensional planning in patients with hip dysplasia (p = 0.004, OR = 7.143) and excessive femoral anteversion (p = 0.012, OR = 1.052), meanwhile the failure risk of stem side two-dimensional planning increased as patients got older (p = 0.003, OR = 1.118). The accuracy of artificial intelligence assisted three-dimensional planning cannot be affected by above factors.
CONCLUSIONS: We confirmed that artificial intelligence assisted three-dimensional preoperative planning showed higher accuracy and stability than two-dimensional preoperative planning in primary cementless total hip arthroplasty. We believe artificial intelligence assisted three-dimensional preoperative planning technology provides surgeons a new reliable choice and offers advantages whether in simple or complicated cases.
METHODS: We conducted a prospective study consisting of 117 consecutive patients who underwent a primary cementless total hip arthroplasty to compare the prediction accuracy of these two methods. The two-dimensional and artificial intelligence assisted three-dimensional planning results of the same patient were compared with the definitive implant size respectively.
RESULTS: The prediction accuracy of artificial intelligence assisted three-dimensional planning for cup and the stem sizes were 66.67% (78/117) and 65.81% (77/117), two-dimensional planning was 30.77% (36/117) and 37.61% (44/117) (p < 0.05). There were poor prediction results of two-dimensional planning in patients with hip dysplasia (p = 0.004, OR = 7.143) and excessive femoral anteversion (p = 0.012, OR = 1.052), meanwhile the failure risk of stem side two-dimensional planning increased as patients got older (p = 0.003, OR = 1.118). The accuracy of artificial intelligence assisted three-dimensional planning cannot be affected by above factors.
CONCLUSIONS: We confirmed that artificial intelligence assisted three-dimensional preoperative planning showed higher accuracy and stability than two-dimensional preoperative planning in primary cementless total hip arthroplasty. We believe artificial intelligence assisted three-dimensional preoperative planning technology provides surgeons a new reliable choice and offers advantages whether in simple or complicated cases.
摘要：
目的：成功的全髋关节置换术依赖于准确的术前计划。然而,常规的术前计划，使用X射线模板的二维方法，已显示出预测组件尺寸的可靠性差。据我们所知,人工智能技术辅助三维术前计划有望提高术前计划的准确性，但缺乏临床证据。因此,在这项研究中，我们比较了这两种演习的预测精度。
方法：我们进行了一项前瞻性研究，该研究由117例连续接受原发性非骨水泥全髋关节置换术的患者组成，以比较这两种方法的预测准确性。将同一患者的二维和人工智能辅助的三维规划结果分别与确定的植入物尺寸进行比较。
结果：人工智能辅助三维规划杯和茎尺寸的预测准确率分别为66.67%(78/117)和65.81%(77/117)，二维计划为30.77%(36/117)和37.61%(44/117)(p<0.05)。髋关节发育不良（p=0.004，OR=7.143）和股骨前倾过度（p=0.012，OR=1.052）患者的二维规划预测结果较差。同时，随着患者年龄的增长，茎侧二维规划的失败风险增加（p=0.003，OR=1.118）。人工智能辅助三维规划的精度不能受到上述因素的影响。
结论：我们证实，在初次非骨水泥全髋关节置换术中，人工智能辅助的三维术前计划显示出比二维术前计划更高的准确性和稳定性。我们相信人工智能辅助的三维术前计划技术为外科医生提供了新的可靠选择，并且无论在简单或复杂的情况下都具有优势。
方法：我们进行了一项前瞻性研究，该研究由117例连续接受原发性非骨水泥全髋关节置换术的患者组成，以比较这两种方法的预测准确性。将同一患者的二维和人工智能辅助的三维规划结果分别与确定的植入物尺寸进行比较。
结果：人工智能辅助三维规划杯和茎尺寸的预测准确率分别为66.67%(78/117)和65.81%(77/117)，二维计划为30.77%(36/117)和37.61%(44/117)(p<0.05)。髋关节发育不良（p=0.004，OR=7.143）和股骨前倾过度（p=0.012，OR=1.052）患者的二维规划预测结果较差。同时，随着患者年龄的增长，茎侧二维规划的失败风险增加（p=0.003，OR=1.118）。人工智能辅助三维规划的精度不能受到上述因素的影响。
结论：我们证实，在初次非骨水泥全髋关节置换术中，人工智能辅助的三维术前计划显示出比二维术前计划更高的准确性和稳定性。我们相信人工智能辅助的三维术前计划技术为外科医生提供了新的可靠选择，并且无论在简单或复杂的情况下都具有优势。
