Clinical application of artificial intelligence-assisted three-dimensional planning in direct anterior approach hip arthroplasty

Abstract

Abstract Purpose The objective of this study was to investigate the efficacy of an artificial intelligence-assisted 3D planning system (AIHIP) in total hip arthroplasty by direct anterior approach and assess the reliability of the AIHIP preoperative program in terms of both interobserver and intraobserver agreement. Methods A retrospective analysis was conducted on patients who underwent unilateral primary THA via direct anterior approach from June 2019 to March 2022. Participants were randomly assigned to receive either the AIHIP system (n = 220) or the 2D template (control group) (n = 220) for preoperative planning. The primary outcome aimed to evaluate the correspondence between the prosthesis selected intro-operation and the one planned preoperatively, as well as to calculate the intraclass correlation coefficient (ICC). Secondary outcomes included operation time, intraoperative blood loss, fluoroscopy times, Harris hip score (HHS), lower limb length difference (LLD), femoral offset (FO), and bilateral femoral offset difference. Results No significant differences were observed in gender, age, body mass index (BMI), aetiology, and American Society of Anesthesiologists (ASA) score between the two groups. Both planning methods exhibited good intraobserver agreement for component planning (ICC: 0.941–0.976). Interobserver agreement for component planning was comparable between the two methods (ICC: 0.882–0.929). In the AIHIP group, the accuracy of acetabular cup and femoral stem prosthetics planning significantly improved, with accuracies within the size range of ± 0 and ± 1 being 76.8% and 90.5% and 79.5% and 95.5%, respectively. All differences between two groups were statistically significant (p < 0.05). Patients receiving AIHIP preoperative planning experienced shorter operation times, reduced intraoperative blood loss, fewer fluoroscopy times, and lower leg length discrepancy (LLD) (p < 0.05). Moreover, they demonstrated a higher Harris hip score (HHS) at three days post-surgery (p < 0.05). However, no significant differences were found in femoral offset (FO), difference of bilateral femoral offsets, and HHS at 1 month after the operation. Conclusion Utilizing AIHIP for preoperative planning of direct anterior approach THA can significantly enhance the accuracy of prosthetic sizing with good reliability, decrease operation time, reduce intraoperative blood loss, and more effectively restore the length of both lower limbs. This approach has greater clinical application value.