Augmented reality-based portable navigation improves accuracy of cup placement in total hip arthroplasty with a short learning curve Authors: Tetsuya Tachibana, Hiroki Katagiri, Takahisa Ogawa, Ryusuke Saito, Tetsuya Jinno Published: 2025 ABSTRACT AbstractBackgroundMalpositioning of the acetabular cup in total hip arthroplasty (THA) can cause complications including dislocation, wear, and aseptic loosening. Although computed tomography (CT)-based navigation systems have improved THA accuracy, the high cost and complexity of these systems have hindered their widespread use. Therefore, this study investigated the accuracy and learning curve of an augmented reality (AR)-based portable navigation system.MethodsThis retrospective study analyzed 106 hips in 87 patients who underwent primary THA using an AR-based portable navigation system between July 2023 and December 2024. A single surgeon performed all surgeries. The accuracy of the intraoperative navigation records for cup inclination and anteversion was assessed by comparing them with postoperative CT-based measurements. Absolute error values exceeding 5° and 10° were defined as outliers > 5° and > 10°, respectively. Receiver operating characteristic (ROC) analysis was used to evaluate the learning curve, and multivariate logistic regression identified risk factors associated with cases classified as outliers (> 5°).ResultsThe mean absolute errors between the navigation records and postoperative measurements were 2.2° for inclination and 3.8° for anteversion. Outliers > 5° were observed in 31% of cases, while outliers > 10° were rare (2.8%). No dislocations or revisions occurred within six months. Learning curve analysis showed consistent accuracy in the initial cases. The only significant predictor of navigation error was intraoperative posterior pelvic tilt change (odds ratio 1.13, p = 0.04). Body mass index and subcutaneous fat thickness were not significantly different.ConclusionsThe AR-based portable navigation provided highly accurate acetabular cup placement without a learning curve. However, excessive intraoperative posterior pelvic tilt increased the risk of anteversion overestimation, highlighting the need for intraoperative pelvic position management.Keywords: Total hip arthroplasty, Augmented reality, Portable navigation, Learning curve, Cup position Background Accurate positioning of the acetabular component in total hip arthroplasty (THA) is critical for optimizing implant longevity and minimizing complications such as dislocation, polyethylene wear, and aseptic loosening [1, 2]. Despite advances in surgical techniques, achieving ideal cup alignment remains challenging, even for experienced surgeons, because of variability in patient anatomy, obesity, intraoperative positioning, and limited visual references [3, 4]. To address this issue, various intraoperative tools have been developed to improve accuracy, including mechanical alignment guides. However, these devices often lack sufficient precision, raising questions regarding their reliability [5]. Computed tomography (CT)-based navigation systems have been introduced to overcome these limitations and have proven to be effective in improving cup placement accuracy [6]. These systems can reduce the incidence of postoperative complications, including dislocations and revisions, by ensuring proper alignment based on preoperative planning and intraoperative feedback [7, 8]. Despite these advantages, CT-based systems are often expensive and time-intensive [6]. Moreover, their complexity necessitates a learning curve for consistent and accurate use [7], limiting their practicality, especially in high-volume centers or resource-constrained settings, where not all orthopedic surgeons can feasibly incorporate them into routine practice. Augmented-reality (AR)-based imageless navigation systems have recently been developed as more accessible alternatives [9]. These systems offer a simplified, low-cost solution that allows real-time intraoperative registration using anatomical landmarks to visualize cup orientation on a smartphone interface [10]. Despite their promise in terms of convenience and efficiency, the actual accuracy of AR-based portable navigation systems in clinical practice remains unclear. Moreover, their learning curves have not yet been adequately characterized. Therefore, the present study (1) evaluated the accuracy of cup placement using an AR-based navigation system with patients in the supine position, (2) determined whether a learning curve exists, and (3) identified the clinical and intraoperative predictive factors associated with cup malalignment using this novel technology. Methods Ethics declaration This retrospective study was approved by the Ethics Committee of Dokkyo Medical University Saitama Medical Center (approval number: 21,070), which waived the requirement for informed consent because the participant data were anonymized and owing to the voluntary nature of the study. All procedures were performed in accordance with the principles of the Declaration of Helsinki. Patients This retrospective study included consecutive patients who underwent primary THA using an AR-based portable navigation system at our institution between July 2023 and December 2024. All procedures were performed by a single orthopedic surgeon who had completed a three-year hip arthroplasty training program and had prior experience with > 100 THA cases. This surgeon had prior experience with over 40 cases using a CT-based navigation system, but no experience with robotic-assisted THA or other imageless portable navigation systems. The inclusion criteria were: (1) primary THA, (2) procedures performed using the supine anterolateral approach, and (3) use of the AR-Hip navigation system (Zimmer Biomet Japan) for acetabular cup placement. The exclusion criteria were: (1) previous hip surgery, (2) high dislocation due to dysplasia of the hip classified as Crowe type III or IV, and (3) intraoperative cup displacement after initial fixation using the AR-based navigation system, such as dislodgement during linear impaction. The same cementless acetabular component (G7 OsseoTi; Zimmer Biomet) was used in all the cases. Patient demographic data, operative data, and records of postoperative dislocation and cup revision within six months after surgery were collected. Surgical procedure and registration of AR-based navigation In all cases, low-dose pelvic CT was performed preoperatively, and three-dimensional planning for acetabular cup placement was conducted using ZedHip software (LEXI Co., Ltd.). The AR-based navigation system itself is an imageless system and does not require CT imaging for its use. The target inclination was set at 40°, and the target anteversion ranged from 15° to 25° based on preoperative planning of the cup for each patient. Acetabular anteversion and excessive femoral anteversion were evaluated on CT images, and changes in pelvic tilt between standing and supine pelvic radiographs were considered to avoid excessive anteversion of the cup in patients with marked standing posterior pelvic tilt. The anterior pelvic plane (APP)-pelvic tilt was defined as the sagittal angle between the APP, formed by the bilateral anterior superior iliac spines (ASIS) and the midpoint of the bilateral pubic tubercles, and the horizontal plane of the operating table [11]. For registration using the AR-based navigation system, the surgeon first inserted two 3-mm bone pins into the iliac crest and attached a clamp-mounted quick reference (QR) code marker (Fig. 1A). Subsequently, three anatomical landmarks, the bilateral ASIS and midpoint of the pubic tubercles, were registered using a pointer equipped with a separate QR code (Fig. 1B). The functional pelvic plane (FPP) was defined as the gravitational vector obtained from the gyroscopic sensor of a smartphone and the line connecting the bilateral ASIS [12]. The APP plane was established by registering the plane formed by the bilateral ASIS and midpoint of the pubic tubercles. This enabled intraoperative recording of the APP-pelvic tilt using an AR-based navigation system. The intraoperative APP-pelvic tilt was recorded during the registration process, immediately after patient positioning on the operating table and before any surgical manipulation. During acetabular cup placement, a smartphone was mounted on the cup impactor, and the real-time radiographic inclination and anteversion angles were displayed by recognizing the QR code on the iliac crest (Fig. 1B). At the final fixation, either at cup impaction or after screw insertion, the angles displayed on the smartphone were recorded as navigation values. All acetabular cup placements were performed using the AR-based navigation system alone, and no fluoroscopic or other imaging assistance was applied. In a limited number of cases, intraoperative radiographs were obtained to evaluate femoral fractures during stem insertion; however, these images were not used to guide or verify acetabular cup positioning. Implementation of an AR-based portable navigation system for acetabular cup placement in total hip arthroplasty. A Preoperative photograph of the right hip of the patient in the supine position. A quick-response (QR) code marker (white arrow) is attached to a pair of bone pins inserted into the right iliac crest. B Registration of the AR-based portable navigation system. Both anterior superior iliac spines and the midpoint of bilateral pubic tuberosity are registered with a pointer of another QR code marker (yellow arrow) using a smartphone. C Cup placement using the AR-based portable navigation system. The smartphone is attached to the cup impactor and recognizes the QR code marker fixed to the iliac crest to show the cup’s radiographic inclination and anteversion angles in real time. AR, augmented-reality; QR, quick-response Accuracy evaluation of AR-based navigation All patients underwent low-dose pelvic CT within one week postoperatively to evaluate potential periprosthetic fractures, with informed consent obtained from each patient. These CT images were analyzed using ZedHip software. A computer-aided design model of the G7 OsseoTi cup was manually adjusted to align with the postoperative multiplanar reconstructed images for accurate assessment of implant positioning (Fig. 2A–C). The radiographic inclination (RI) and radiographic anteversion (RA) angles of the cup were measured using ZedHip software and adjusted to the preoperative FPP coordinate system [7]. Postoperative CT assessment of acetabular cup orientation and soft tissue thickness anterior to the ASIS. A Measurement of the acetabular cup’s alignment in the coronal plane of CT images. B Sagittal plane. C Axial plane. D Axial view showing measurement of soft tissue thickness in front of bilateral ASIS. Measurements are taken bilaterally (yellow arrows), and the average of two sides is used. ASIS, anterior superior iliac spine; CT, computed tomography To evaluate the accuracy of the AR-based navigation system, the differences and absolute differences between the intraoperative navigation-recorded values and postoperative CT-based measurements were calculated. Cases in which the absolute differences between intraoperative and postoperative values exceeded 5° or 10° were defined as > 5° and > 10°, respectively [13, 14]. Correlation of outliers in cup alignment and clinical parameters To investigate the potential factors influencing cup alignment accuracy, we retrospectively collected patient data from electronic medical records, including patient age, sex, body mass index (BMI), and diagnosis, as well as the number of cases in which the surgeon had used the AR-based navigation system. As an intraoperative parameter, we measured the change in pelvic tilt (ΔAPP-pelvic tilt), defined as preoperative minus intraoperative APP-pelvic tilt. In this parameter, positive values indicated intraoperative posterior tilt compared with the preoperative position. In addition, to assess the registration-related factors in the AR-based navigation system, we evaluated the subcutaneous fat thickness in the ASIS on preoperative CT scans (Fig. 2D), defined as the “soft tissue thickness in front of the ASIS.” Measurements were taken bilaterally, and the average of the two sides was used for analysis. Statistical analysis All statistical analyses were performed using STATA version 16 (StataCorp LP, College Station, TX, USA). Continuous variables are presented as means ± SD, and p < 0.05 was considered statistically significant. The normality of distributions was assessed using the Shapiro–Wilk test and histograms. For normally and non-normally distributed variables, unpaired Student’s t-tests and Mann–Whitney U test, respectively, were applied. Categorical variables were analyzed using the chi-square test. To assess the intra-observer reliability of the AR navigation system, the surgeon performed the registration procedure twice for 50 hips and measured the intraoperative APP-pelvic tilt in each instance. The intra-observer intraclass correlation coefficient (ICC) for intraoperative APP-pelvic tilt measured using the AR-based navigation system was 0.89° (95% CI: 0.81–0.93, p < 0.01). The minimal detectable change was calculated as standard deviations (SD)*√ 1-ICC * 1.96 *√ 2 [15, 16]. The minimal detectable change in the intraoperative APP-PT measurement was 3.1°. Receiver operating characteristic (ROC) curve analysis was performed to determine the threshold number of cases required to achieve proficiency in cup placement using the AR-based navigation system. The outlier incidence > 5° was calculated in increments of five cases. For each five-case grouping, the sensitivity and specificity were computed using a ROC curve, and the area under the curve (AUC) was assessed. To further evaluate the presence of a learning curve, we performed a continuous cumulative sum (CUSUM) analysis for the absolute errors of radiographic inclination and anteversion measured by the AR-based navigation system. The threshold for acceptable error was set to 5°, and the CUSUM value for each case was calculated as the cumulative sum of the difference between the absolute error and the threshold. A downward trend in the CUSUM plot indicate that the accuracy was maintained below the acceptable error, whereas an upward trend suggested deterioration. Multivariate logistic regression analysis was performed to identify the risk factors associated with outliers > 5°. The a priori selected variables included age, sex, BMI, preoperative pelvic tilt, diagnosis, number of AR-based navigation-assisted THA procedures performed by the surgeon, ΔAPP-pelvic tilt, and soft tissue thickness in front of the ASIS. Initially, univariate logistic regression analysis was conducted for each variable. Those with p ≤ 0.20 were included in the final multivariate model. The final multivariate analysis included age, sex, BMI, soft tissue thickness in front of the ASIS, and ΔAPP-pelvic tilt. Additionally, scatterplots were generated, and linear regression was performed to examine the relationship between errors in RA-defined intraoperative navigation values minus postoperative measurements and ΔAPP-pelvic tilt. A positive RA error indicated that the navigation system had overestimated the cup anteversion compared with the actual postoperative value. Results Patient demographics and intraoperative findings After excluding one case with obvious cup translation during liner impaction observed intraoperatively and one case with a history of open reduction and internal fixation for pelvic fracture, the final analysis included 106 hips from 87 patients. No cases of high dislocation classified as Crowe type III or IV were included during the study period. The demographic and operative data of the 87 patients are shown in Table 1. The mean preoperative APP-pelvic tilt angle was 1.6° ± 7.2 (range: − 16.6° to 19.6°). The mean soft tissue thickness in front of the ASIS was 30.9 ± 20.8 mm (range: 3.8–102 mm). The average operative time was 93.8 ± 19.6 min (range: 57–175 min) and the mean estimated blood loss was 196.4 ± 95.8 mL (range: 50–600 mL). The mean intraoperative APP-pelvic tilt angle was − 12.1° ± 6.5 (range: − 29.1° to 6.3°). The ΔAPP-pelvic tilt (preoperative – intraoperative value) was 13.6° ± 6.5 (range: − 3.3° to 29°), with 97% of cases showing positive values, indicating the posterior pelvic tilt during surgery compared with the preoperative position. No dislocation or cup revision was observed. Table 1Patient demographic and clinical characteristicsPatient characteristics87 patients 106 hipsAge, years68.9 ± 9.8Sex, female/male (%)85 (79) /22 (21)Body mass index, (kg/m2)25.7 ± 4.8Diagnosis, n (%) Primary hip osteoarthritis36 (34) Dysplasia62 (58) Osteonecrosis of the femoral head3 (3) Other5 (5)Preoperative pelvic tilt, (°)1.6 ± 7.2Soft tissue thickness in front of the ASIS, (mm)30.9 ± 20.8Postoperative pelvic tilt, (°)1.5 ± 6.8Intraoperative pelvic tilt, (°)-12.1 ± 6.5Duration of surgery, minutes93.8 ± 19.6Blood loss, mL196.4 ± 95.8 Patient demographic and clinical characteristics Accuracy of navigation for cup inclination and anteversion The intraoperative navigation-recorded RI and RA angles were 40.2° ± 1.7° (range: 35.0°–44.4°) and 20.6° ± 3.1° (range, 12.8°–28.0°), respectively. The preoperatively planned target angles were 40.0° ± 0° for inclination and 19.6° ± 2.6° (range: 15.0°–25.0°) for anteversion. The absolute error between the navigation-recorded values and the postoperative CT measurements was 2.2° ± 1.8° (range: 0°–8.5°) for RI and 3.8° ± 2.6° (range: 0.1°–16.4°) for RA. Compared with the target values, the absolute error was 2.1° ± 1.7° (range: 0°–7.0°) for RI and 3.3° ± 2.5° (range: 0°–15.8°) for RA (Table 2). The errors between navigation-recorded values and postoperative measurements were 0.1° ± 2.8° (range: − 6.0° to 7.0°) for RI and 3.1° ± 3.5° (range: − 7.7° to 15.8°) for RA. The scatterplots of the errors are shown in Fig. 3. Notably, 81% of the RA errors were positive, indicating that the navigation system tended to overestimate anteversion relative to the actual postoperative measurements. Overall, 69% of cases were within 5° and 97% were within 10° of the postoperative CT measurements for both radiographic inclination and anteversion. Table 2Angle differences between AR-guided intraoperative records and postoperative CT or targeted anglesParameterNavigation-postoperative measurementsNavigation-targeted anglesRadiographic inclination (º)2.2 ± 1.82.1 ± 1.7Radiographic anteversion (º)3.8 ± 2.63.3 ± 2.5AR Augmented reality, CT Computed tomography Angle differences between AR-guided intraoperative records and postoperative CT or targeted angles AR Augmented reality, CT Computed tomography Scatterplots of errors in radiographic inclination and anteversion between intraoperative navigation records and postoperative measurements. Positive values indicate an overestimation of the cup angles by the navigation system compared with postoperative CT-based measurements. Notably, 81% of cases show positive RA errors, suggesting that the AR-based navigation system tends to overestimate radiographic anteversion relative to the actual cup position. CT, computed tomography; RA, radiographic anteversion; AR, augmented reality Outlier and learning curve analyses A total of 33 hips (31%) demonstrated outliers > 5°, including nine hips (8.5%) for RI and 29 hips (27%) for RA. Outliers > 10° were found in only three hips (2.8%), all of which involved the RA; no outliers > 10° were observed for RI. To evaluate the learning curve, ROC curves and AUC values were analyzed in five-case increments (Fig. 4). The ROC curve closely followed the diagonal reference line, and the AUC values remained at approximately 0.5 throughout the sequence of cases. To further evaluate the learning curve, CUSUM analysis of the absolute errors in radiographic inclination and anteversion, using a 5° threshold, demonstrated a consistent downward trend for both parameters throughout the case sequence (Fig. 5). These findings indicated that the accuracy of cup placement using the AR navigation system was maintained within the acceptable error from the initial cases; thus, we observed no apparent learning curve associated with the use of this system. Operator learning curve for cup placement using AR-based navigation system. A ROC curve analysis to identify the learning curve threshold for proficiency in cup placement for every five cases. The ROC curve evaluates the occurrence of outliers > 5 in cup alignment and closely follows a 45-degree diagonal line. B AUC values for every five-case increment showing that the AUC for the initial five cases has already approached 0.5, suggesting no apparent learning curve in the AR-based portable navigation system. ROC, receiver operating characteristic; AUC, area under the curve; AR, augmented reality CUSUM plots of absolute errors in radiographic inclination and anteversion using a 5° threshold. CUSUM, cumulative sum Factors predicting inaccurate cup placement The results of the final multivariate logistic regression analyses are presented in Table 3. Soft tissue thickness in front of the ASIS and BMI were not significant risk factors. The only independent predictor of inaccurate cup positioning > 5° was the ΔAPP-pelvic tilt between the preoperative and intraoperative measurements, with an odds ratio of 1.13 (p = 0.04). Scatterplots illustrating the relationship between ΔAPP-pelvic tilt and RA errors are shown in Fig. 6. Linear regression analysis revealed a significant positive correlation between ΔAPP-pelvic tilt and RA error (β = 0.32, p < 0.001, R² = 0.35), indicating that a greater intraoperative posterior tilt of the pelvis was associated with an increased overestimation of the RA by the AR-based navigation system. Specifically, for every 1° increase in intraoperative posterior APP-pelvic tilt, the RA value displayed by the AR-based navigation system was overestimated by approximately 0.3°. In contrast, linear regression analysis revealed no significant association between ΔAPP-pelvic tilt and RI errors (β = 0.04, p = 0.34, R² = 0.01). The coefficient of determination was extremely low (R² = 0.01), indicating that ΔAPP-pelvic tilt explained very little of the variation in RI errors. Table 3Risk factors for outlier > 5° in acetabular cup placement using the AR-based navigation systemParameterOutlier > 5° *p-valueAge1.000.86Sex (female)0.490.24Body mass Index0.990.95Soft tissue thickness in front of the ASIS1.020.60ΔAPP-pelvic tilt1.13 0.040 Data are expressed as odds ratios. Odds ratios were calculated using multivariate regression analysis adjusted for age, sex, body mass index, soft tissue thickness in front of the ASIS, and ΔAPP-pelvic tilt Risk factors for outlier > 5° in acetabular cup placement using the AR-based navigation system Data are expressed as odds ratios. Odds ratios were calculated using multivariate regression analysis adjusted for age, sex, body mass index, soft tissue thickness in front of the ASIS, and ΔAPP-pelvic tilt Scatterplots illustrating the relationship between APP-pelvic tilt changes and errors in RA in AR navigation. The significant positive correlation (β = 0.32, P < 0.001, R² = 0.35) indicates that greater intraoperative posterior pelvic tilt leads to RA overestimation. Specifically, for every 1° intraoperative increase in posterior APP-pelvic tilt, the RA indicated by navigation is overestimated by approximately 0.3°. APP, anterior pelvic plane; RA, radiographic anteversion; AR, augmented reality Discussion This study evaluated the accuracy of acetabular cup placement using an augmented reality-based portable navigation system, assessed the presence of a learning curve, and identified clinical and intraoperative factors associated with cup malalignment. The key finding was that the AR-based navigation system achieved a high level of accuracy in acetabular cup placement, even in obese patients and initial cases. Although several previous studies have examined AR-based navigation [9, 10], the present study is unique in that it simultaneously analyzed accuracy, learning curve characteristics, and intraoperative predictors within a single AR-based platform. The mean absolute errors in radiographic inclination and anteversion were 2.2° and 3.8°, respectively, with outliers > 10° observed in 0% and 3% of the inclination and anteversion cases, respectively. Importantly, multivariate logistic regression identified a change in the APP-pelvic tilt from the preoperative to the intraoperative position as a significant risk factor for inaccuracy. In the AR navigation system, each 1° increase in intraoperative posterior APP-pelvic tilt compared with the preoperative value led to an RA overestimation by approximately 0.3°. These results provided a clinically actionable insight into how intraoperative pelvic tilt directly influences the accuracy of the AR navigation system. Previous studies evaluating the accuracy of portable navigation systems reported absolute differences between intraoperative navigation records and postoperative measurements of 2.7°–3.3° for inclination and 2.7°–3.8° for anteversion [17, 18], consistent with the findings in the present study. Hayashi et al. identified obesity as a risk factor for the reduced accuracy of accelerometer-based navigation systems [18], contrary to the multivariate analysis findings in the present study. Similarly, Richolt et al. reported that the thickness of the subcutaneous fat anterior to the ASIS and pubic tubercle (measured by ultrasound) could compromise registration accuracy [19]. However, in our study, soft tissue thickness in front of the ASIS was not significantly associated with cup malalignment. One possible explanation is the registration method used in our AR-based navigation system, which involved lightweight QR code markers and smartphone-based pointing devices. This setup allowed manual displacement of the overlying soft tissue thickness in front of the ASIS during landmark registration, potentially preserving accuracy even in obese patients. Although CT-based navigation systems are known for their high accuracy in cup placement, previous studies have reported substantial learning curves [7]. In contrast, our study, which employed a detailed analysis using ROC curves and AUC values, demonstrated a consistently low incidence of outliers in the initial five cases using the AR-based navigation system, indicating no apparent learning curve. One possible reason for this is the simplicity of the registration process. Whereas CT-based systems require multiple steps and extensive data input, the AR-based navigation system uses only three anatomical landmarks: the bilateral ASIS and midpoint of the pubic tubercles. This streamlined registration process may facilitate the rapid acquisition of the technique, even for surgeons less experienced in the AR-based navigation system. Nevertheless, this study does not intend to negate the value of CT-based navigation systems. While the AR-based system is useful for controlling cup orientation, it does not provide three-dimensional control of cup position. In our series, one case experienced cup displacement during liner impaction, likely due to insufficient medialization of the cup after cautious impaction in osteoporotic acetabular bone in a patient with developmental dysplasia of the hip. Therefore, CT-based navigation may remain advantageous for precise three-dimensional positioning of the cup. Several studies have reported that for every 1° increase in posterior APP-pelvic tilt, RA decreases by approximately 0.7°–0.8° [20, 21], compared with approximately 0.3° in the present study. The smaller degree of RA in this study change may be attributed to the methodological difference between the intraoperative measurement of pelvic tilt using the AR-based navigation system and preoperative measurement based on CT image. Nevertheless, because the AR system allows real-time measurement of pelvic tilt, surgeons should be cautious in cases of a marked posterior tilt during surgery. Intraoperative APP-pelvic tilt was recorded during registration before any surgical manipulation, minimizing the influence of limb traction or retractors. The posterior pelvic tilt observed in most cases was considered likely to result from anesthesia-related muscle relaxation and differences in support between the CT and operating tables. This study has several limitations. First, it focused on procedures performed by a single hip surgeon, which may have limited the generalizability of the findings. Although the AR-based navigation system features a simplified registration process that may facilitate adoption by less-experienced surgeons, procedural accuracy and learning curves could differ among surgeons with varying backgrounds and levels of experience. In addition, the operating surgeon had prior experience with more than 40 cases using a CT-based navigation system. This background may have enhanced familiarity with anatomical registration; however, its influence on the accuracy observed in this study is likely limited, as the AR-based workflow relies only on simple landmark registration without CT-image merging or intraoperative matching. Second, the sample size was relatively small, which potentially reduced the statistical power of the analysis. Well-designed multicenter studies involving multiple surgeons are required to evaluate the effectiveness of AR-based portable navigation systems. Third, the follow-up period in this study was limited to 6 months. Longer-term follow-up is required to confirm the durability of cup positioning and its long-term safety. Fourth, the AR-based navigation system used in this study is designed solely for acetabular component orientation and does not provide intraoperative assessment of femoral offset or limb length. This is because the current platform does not incorporate femoral reference markers. Future system upgrades integrating femoral landmarks may enable comprehensive three-dimensional assessment of both acetabular and femoral parameters. Fifth, previous studies have demonstrated that AR technology combined with CT-based navigation can achieve high accuracy in cup positioning [22]. While the present study focused on an imageless AR system, acknowledging these hybrid AR–CT approaches provides important context and suggests that future integration of CT-based data may further enhance navigation precision. Finally, the intraoperative APP-pelvic tilt was measured using the AR navigation system, whereas the preoperative APP-pelvic tilt was assessed using CT data. Because these two methods are not identical, differences in measurement techniques may have influenced the results. A standardized approach for assessing intraoperative pelvic tilt equivalent to preoperative CT-based measurements could enhance the accuracy of AR-based navigation. Other intraoperative factors such as anesthesia depth or operating table support may also affect pelvic tilt. However, these variables were outside the primary scope of this study. Future prospective research is needed to evaluate their influence. Conclusions This is the first study to evaluate both the accuracy and learning curve of an AR-based navigation system for acetabular cup placement. The results demonstrated that the AR-based navigation system achieved high accuracy in radiographic inclination, with over 90% of cases showing an error of less than 5°. However, changes in pelvic tilt during surgery were a key risk factor. Specifically, substantial intraoperative posterior pelvic tilt may lead to an overestimation of radiographic anteversion by the AR-based navigation system. Surgeons should be aware of this potential source of error and interpret the navigation values with caution in such cases. Acknowledgements This study was supported by the JSPS KAKENHI (grant number JP 23K15694). We thank Editage (www.editage.jp) for the English language editing. Abbreviations Anterior pelvic plane Augmented reality Anterior superior iliac spine Area under the curve Body mass index Confidence interval Computed tomography Cumulative sum Change in anterior pelvic plane tilt (preoperative – intraoperative) Functional pelvic plane Intraclass correlation coefficient Quick response (code) Radiographic anteversion Radiographic inclination Receiver operating characteristic Standard deviation Total hip arthroplasty Authors’ contributions TT, HK, and JT contributed to the study conception. TO contributed to the statistical analysis. RS supervised the study. All the authors reviewed the draft of the manuscript and critically revised it for intellectual content. All authors have approved the final version of the manuscript for publication. Funding None. Data availability The datasets generated and/or analyzed in the current study are available from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate This retrospective study was approved by the Ethics Committee of Dokkyo Medical University Saitama Medical Center (approval number: 21,070), and all procedures were performed in accordance with the principles of the Declaration of Helsinki. Because the participant data were anonymized, and because of the voluntary nature of the study, the requirement for informed consent was waived by the Ethics Committee of Dokkyo Medical University Saitama Medical Center. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The datasets generated and/or analyzed in the current study are available from the corresponding author upon reasonable request.