Optimized treatment of hip ankylosis in spondylitis with direct anterior approach arthroplasty using AI navigation: a case report Authors: Jiaxin Dong, Zhiming Cheng, Zaiyang Liu, Zhonghua Xu, Yuan Zhang Published: 2026 ABSTRACT BackgroundComplex hip-spine syndrome combined with hip ankylosis is clinically challenging to manage. Traditional surgeries are associated with insufficient precision and significant trauma, which easily lead to poor lumbar-pelvic-hip alignment and affect prognosis. With the development of digitalized and intelligentized surgical techniques, the integration of artificial intelligence (AI), computer navigation, and minimally invasive techniques has provided a new solution for such complex cases, enabling optimized surgical outcomes through precise planning and operation in line with the damage-control philosophy.Case DescriptionA young male patient with severe hip-spine deformity and hip ankylosis underwent treatment. The first-stage hip surgery was performed using AI, computer navigation and minimally invasive total hip arthroplasty. The second-stage scoliosis correction was conducted based on simulation planning with a three dimensional (3D) printed model. At the 30-month postoperative follow-up, the Harris Hip Score (HHS), Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), and 36-Item Short-Form Health Survey (SF-36) scores increased by 75, 143, and 350 points, respectively. The patient’s overall quality of life was significantly improved. Spinal function parameters including cervical balance vertical axis (CBVA), Scoliosis Research Society-22 questionnaire (SRS-22), and Oswestry Disability Index (ODI) were also significantly enhanced. Radiological examinations (anteroposterior pelvic and standing full-length X-rays) showed that the inclination and anteversion of the acetabular component were consistent with AI’s preoperative predictions, reflecting favorable lumbar-pelvic-hip adaptation.ConclusionsIn the treatment of hip ankylosis in complex hip-spine syndrome, the application of AI combined with navigation technology can significantly improve surgical precision, reduce trauma, promote coordinated lumbar-pelvic-hip alignment, and effectively enhance patients’ joint/spinal function and quality of life. It provides a new digitalized and intelligentized solution for the surgical management of such complex cases. Highlight box Key findings • The staged surgical strategy (first-stage total hip arthroplasty (THA), second-stage spinal corrective surgery) integrating artificial intelligence (AI) planning, computer navigation, and minimally invasive direct anterior approach (DAA) successfully resolves the intraoperative positioning dilemma of AS patients unable to lie prone due to bilateral hip ankylosis at 45° flexion. What is known and what is new? • Ankylosing spondylitis (AS) creates complex hip-spine deformities where pelvic tilt critically affects acetabular orientation and prosthesis stability. Traditional THA often carries high trauma and failure risks, while the minimally invasive DAA remains highly sensitive to these spinal-pelvic biomechanics. • This study reports the first integration of AI preoperative planning, computer navigation, and DAA-THA to address these challenges. By using AI to simulate post-correction pelvic positioning, the study overcomes limitations in predicting dynamic changes. It verifies the durability of this “AI planning + navigation + staged surgery” strategy through 30 months of follow-up, providing key clinical evidence for managing complex AS cases. What is the implication, and what should change now? • This study establishes a safe and precise digital treatment paradigm for AS patients with severe spinal-pelvic-hip malalignment and positioning limitations. By integrating AI-driven prediction of pelvic functional positions with computer navigation and staged surgery, the approach overcomes traditional technical bottlenecks and validates the clinical value of combining digital technologies with damage-control principles. Future efforts will focus on expanding cohort sizes to verify generalizability, promoting clinical adoption for complex hip pathologies, and continuously optimizing AI algorithms based on long-term data to enhance dynamic prediction accuracy and personalized surgical planning. Introduction Background Ankylosing spondylitis (AS) is a complicated musculoskeletal disorder affecting both spinal vertebra and synovial joints. Among its multiple pathologies, the hip-spine deformity presents an extreme challenge in orthopedic reconstructive surgery. In these patients, the normal hip-spine adaption mechanism is impaired due to the anteflexion of the lumbar vertebra and retroversion of the pelvis. Although total hip arthroplasty (THA) has been reported to be effective in the treatment of end-stage AS, the conventional technique is also associated with significant problems, including notable operative trauma, complications, and adverse outcomes. Specifically, component impingement and bearing wear during normal hip motion have raised great concern regarding early prosthesis failure (1). In recent years, artificial intelligence (AI) has rapidly advanced into clinical decision-making and surgical assistance. AI application in joint surgery facilitates the digitization, precision, and intelligentization of surgical procedures, and provides surgeons with an approach to implement rapid prototype technology and patient-specific instrumentation. Rationale and knowledge gap In a previous study, we reported that robot-assisted total knee arthroplasty (TKA), compared to traditional manual TKA, significantly reduced operative trauma, inflammatory response, and the incidence of mechanical errors, while promoting postoperative comfort and functional recovery (2). Computer navigation has been utilized in our work for several years and plays a significant role in primary THA (3). However, despite the proven value of AI and computer navigation in conventional joint arthroplasty, their potential benefits in complex THA scenarios—particularly for hip-spine syndrome caused by AS pathology—remain to be verified. Conventional THA techniques are inadequate for such complex cases, highlighting an unmet need for precise and intelligent surgical solutions to address the unique challenges of AS-related hip-spine deformity. Objective The objective of this study is to present an innovative program integrating AI-assisted preoperative planning and computer navigation-assisted surgery, and to demonstrate its precise and intelligent efficacy in the management of complex hip-spine syndrome resulting from AS pathology. We present this article in accordance with the CARE reporting checklist (available at https://aoj.amegroups.com/article/view/10.21037/aoj-2025-1-94/rc). Case presentation All procedures performed in this study were in accordance with the ethical standards of the institutional and/or national research committee(s) and with the Declaration of Helsinki and its subsequent amendments. Written informed consent was obtained from the patient for the publication of this case report and accompanying images. A copy of the written consent is available for review by the editorial office of this journal. A 33-year-old man complained of low-back pain and disability in hip motion for over 20 years. Several anti-inflammatory drugs such as diclofenac sodium and sulfasalazine were administrated intermittently for 10 years, which had been ineffective in the recent 2 years. The patient had a 144° overall razor back deformity in shape. He could not look forward horizontally and lie down on his back. The patient had an obvious loss of range of motion (ROM) of lumbar vertebra and both hips. His knee and ankle function were normal, with a ROM of 0–130° for the knees, and 7–8 dorsiflexion of 20° to plantarflexion of 45° for the ankles. His walking mainly depended on over-flexion and over-extension activities of the tibiotalar and tarsometatarsal joints. Blood tests showed mild anemia (105 g/L of hemoglobin), erythrocyte sedimentation rate of 2 mm/h, C-reactive protein of 12.2 mg/L, positive result for human leukocyte antigen (HLA) B27, and negative results for rheumatoid markers and antinuclear antibody profile. Full-length spine X-ray images demonstrated classical bamboo-like changes of the vertebra, with severe scoliosis and kyphosis deformities (Figure 1A,1B). Sagittal pelvic X-ray images showed extreme retroversion of the pelvis, total ankylosis of both sacroiliac joints (grade 4), and bilateral hip ankylosis at 45° of flexion, 25° of external rotation on the right, and 30° of internal rotation on the left (Figure 1C). In the spinal three dimensional (3D) image, objective spinal and pelvic measurements were as follows: lumbar lordosis (LL) was measured at −11.9°, thoracic kyphosis (TK) at 50.1°, sagittal vertical axis (SVA) at 41.7 cm, total pelvic angle (TPA) at 83.5°, pelvic incidence (PI) at 64.3°, pelvic tilt (PT) at 44.4°, and sacral slope (SS) at 19.9°. These parameters are standard radiographic assessments used to evaluate the patient’s spinal alignment in the sagittal plane (Figure 1D). According to the modified New York criteria, the patient was diagnosed with AS combined with scoliosis and kyphosis, as well as bilateral hip ankylosis. Preoperative radiographs. (A,B) Anteroposterior X-rays of the spine; (C) full-length weight-bearing X-ray image of lower extremities; (D) 3D reconstruction of computed tomography data showing lumbo-pelvic-hip alignment, and main spinal-pelvic parameters. 3D, three dimensional; CL, cervical lordosis; LL, lumbar lordosis; PI, pelvic incidence; PT, pelvic tilt; SS, sacral slope; SVA, sagittal vertical axis; TK, thoracic kyphosis; TPA, total pelvic angle. Opening wedge osteotomy, target soft tissue balancing, and THA were planned for the first stage. After partial recovery of hip functions, corrective surgery for the spine was considered for the second stage. Preoperative plan Materialise Mimics (version 21.0, Materialise, Belgium) and Surgimap (version 2.2.15.5, Nemaris Inc., USA) were employed to reconstruct the full-length spine computed tomography (CT) 3D view, to design the parameters of preoperative simulation of spinal osteotomy, and to calculate the angles for scoliosis and kyphosis correction. An estimated correction of 18.8° would be made on the pelvic tilt after pedicle subtraction osteotomy (an angle of 36° at L3, Figures 2,3). Decision-making for spine and hip surgery. (A,B) Prediction of pelvic functional position adapted to THA based on the preoperative simulation of spine osteotomy on a 3D model. (C,D) Deduction of pelvic orientation and prosthesis position under pathological conditions based on the pelvis functional position. (E,F) Anteroposterior X-ray images of the pelvis before and immediately after THA. 3D, three dimensional; PI, pelvic incidence; PT, pelvic tilt; THA, total hip arthroplasty. Temporal changes of lumbo-pelvic-hip alignment at different time intervals after spine surgery. (A) Anteroposterior X-ray of the spine and pelvis immediately after THA and before spine surgery; (B) immediately after spine surgery; (C) 18 months after spine surgery, full-length weight-bearing lower extremities X-ray 18 months after spine surgery; (D) main spinal-pelvic parameters at the last follow-up. CL, cervical lordosis; LL, lumbar lordosis; PI, pelvic incidence; PT, pelvic tilt; SS, sacral slope; SVA, sagittal vertical axis; THA, total hip arthroplasty; TK, thoracic kyphosis; TPA, total pelvic angle. Following spine orthopedic surgery, the 3D morphology of the pelvis and hips was reconstructed with the AI-HIP software (Version 3.0, Longwood Valley Technology, China) is a deep learning-based preoperative planning platform. Its algorithmic core consists of an Attention U-Net with a PointRend module, which automates the segmentation of pelvic and femoral CT images. Following segmentation, a Stacked Hourglass Network is employed to identify key anatomic landmarks, including the anterior superior iliac spines (ASIS) and the center of the femoral head, enabling automatic pelvis correction and deformity assessment. The system was trained on a dataset comprising over 1.2 million CT images from 3,000 patients. According to the validation study by Chen et al. [2022], this algorithm achieves a Dice similarity coefficient (DSC) of 0.973, with an average processing time of approximately 1.86 minutes per case. This high level of automation and accuracy allowed us to reliably simulate the dynamic ’spine-pelvis-hip’ relationship in this complex ankylosis case (4). The lumbar-pelvic-hip alignment at the endpoint of therapy was simulated on the basis of the pelvic anteversion and rotation obtained by spinal osteotomy to acquire the abduction angle of a standard implanted prosthesis. The acetabular anteversion angle was predicted and adjusted through anterior and posterior impingement animation. The pelvis was then restored to the preoperative state to determine the orientation of the prosthesis implanted in this hip surgery (the prosthesis was preoperatively planned to be implanted at an anteversion angle of 0–3° and angle of 50–55°, Figure 2). First-stage surgery (April, 2021) Computer navigation assisted bilateral opening wedge osteotomy, targeted soft tissue balancing, and THA were performed via direct anterior approach (DAA). The patient was positioned supine in a beach chair shape on a standard operating table to accommodate complex hip-spine deformity. Intraoperatively, conventional DAA exposure revealed complete synostosis between the proximal femur and acetabulum, and eliminating surgical interspace. Femoral neck osteotomy levels were pre-determined via AI-HIP prediction based on the saddle zone and superior lesser trochanter margin, then verified intraoperatively by the acetabular lateral superior margin remnant under StealthStation 7.0 (Medtronic, USA) navigation guidance. Registration was completed using the native acetabular anteroposterior margin remnant and ossified glenoid labrum. Guided by navigation, acetabular reaming was performed at the anatomical center of rotation; autologous femoral head impaction grafting addressed acetabular offset deficiency secondary to disuse osteoporosis and acetabular protrusion. A 52 mm porous cementless acetabular shell (Pinnacle Gription, DePuy, USA) was implanted at 0° anteversion and 50° inclination (Figure 2E,2F). Femoral preparation included targeted piriformis and posteromedial capsule release, followed by implantation of a size 10 fully coated tapered femoral stem (Corail, DePuy, USA) using standard techniques. Computer navigation enabled precise optimization of stem height, femoral head length, hip ROM, combined offset, and leg length to achieve soft tissue tension balance and joint stability. Post-THA, left hip kinematics were compromised by pelvic rotational deformity-induced femoral adduction/internal rotation malalignment; this caused minor trochanter-ischial tuberosity impingement at 5° adduction and spontaneous anterior dislocation in neutral position. Impingement-related bony prominences were resected, partially resolving instability, though dislocation risk persisted with >10° adduction. The operative time was 4.5 hours, and the blood loss was 1,200 mL. Immediate postoperative management included partial weight-bearing with two crutches and active hip muscle exercises, followed by 4 weeks of strict hip postural restriction (≤10° internal/external rotation; ≤90° flexion). Second-stage surgery (May, 2021) Corrective vertebral osteotomy and lumbar fusion in the prone position were performed when the active flexion reached 50°, and the hip abductor strength was grade 3–4, which indicated the best opportunity for the spine surgery. Pedicle screws placed bilaterally at T10-L2 and L4-S1 and connected by conventional rods, and satellite rods were placed between T12 and L4. Pedicle subtraction osteotomy was performed at L3 to correct kyphosis and restore the lumbar lordosis by 30–35°. Ponte osteotomy at L1 and asymmetric osteotomy at L1-2 through the intervertebral space were conducted to restore the lumbar lordosis by 15–20°. Asymmetric osteotomy was performed at L2 to correct coronal scoliosis, and the L1-2 intervertebral disc, part of the pedicle at L2, the L2 superior endplate and partial vertebral body were removed (Figure 3). Intraoperative spinal cord function was monitored by combined motor potential and sensory potential. The surgery lasted 11 hours, with an estimated blood loss of 220 ml. The patient was bedridden for 1 week and wore a thoracolumbosacral orthosis brace for 3 months. Clinical outcome The clinical results were obtained by continuous follow-up visits at 1, 3, 6, 12, 18 and 30 months after spine surgery. The patient was 1.35 m tall before surgery, 1.45 m after THA, and 1.73 m after spinal surgery. No change in his body weight (60–65 kg) was observed. He was satisfied with significantly improved visual field, quality of life, and enhanced self-confidence in social activities (Figure 4A-4C). A walking aid was used for 3 months after spine surgery, and his continuous walking distance increased from 50 m in the first month to 100 m in the third month. At 6 months after surgery, his independent walking distance was approximately 200 m. From one year after surgery to the end of the follow-up, the patient could take part in outdoor activities and walk up-down stairs freely. Following surgery. (A,B) General appearance, weight-bearing, trunk alignment; (C) visual field changes before hip surgery and at the end of follow-up; (D,E) temporal changes in hip function score and acetabular position during continuous follow-up visits. HHS, Harris Hip Score; L, left; R, right; SF-36, 36-Item Short-Form Health Survey; THA, total hip arthroplasty; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index. The patient’s self-assessment results showed great satisfaction in terms of the surgery. By the end of the follow-up, the patient showed significant improvements in hip function, osteoarthritis symptoms, and overall quality of life. Specifically, the Harris Hip Score (HHS)—which assesses hip function (where a higher score indicates better function)—improved from 8 to 83 points. The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC)—which measures the severity of osteoarthritis (where a lower score indicates fewer symptoms)—decreased from 230 to 87 points. Additionally, the 36-Item Short-Form Health Survey (SF-36) score—a measure of overall health status (where a higher score reflects a better quality of life)—increased from 182.8 to 572.1 points (Figure 4D). The patient’s postoperative imaging revealed improvements in key spinal and pelvic parameters. TK was notably reduced from 50.1° to 30.8°, indicating a positive correction of thoracic curvature. PT showed a significant reduction from 44.4° to 25.6°, which may improve spinal and pelvic alignment and reduce discomfort. TPA also decreased, reflecting an overall improvement in pelvic orientation. LL transitioned positively from −11.9° to 26.4°, suggesting a restored lumbar curvature. SS increased from 19.9° to 38.7°, which can indicate a more optimal sacral alignment contributing to better spinal balance. Additionally, the SVA improved substantially from 41.7 to 10.3 cm, which may enhance stability and balance. PI remained unchanged at 64.3°, providing a stable reference for other parameter changes (Figure 3C,3D, Table 1). The value of spinal function parameters including chin-brow vertical angle, by the end of the follow-up, the patient’s spinal alignment, quality of life, and functional disability showed significant improvement. Specifically, the spinal curvature (Cobb angle) was corrected from 81.7° to 26.6°. The Scoliosis Research Society-22 (SRS-22) score—which assesses health-related quality of life in scoliosis patients (where a higher score is better)—improved from 67 to 71 points. Meanwhile, the Oswestry Disability Index (ODI)—which measures the degree of disability caused by low back pain (where a lower percentage indicates less disability)—decreased from 42.22% to 22.20%. Anteroposterior pelvic and full-length lower limb X-ray images showed the position of the acetabular component and leg length change as predicted by AI, which reflects adaptive recovery in the lower lumbar–pelvic–hip complex, also in line with the preoperative assessment for spinal osteotomy (Figures 3B,3C,4E). The final inclination/anteversion of the acetabular components was 37.6°/7.2° on the left hip, and 35.3°/9.5° on the right hip at the latest follow-up, all of which were within the Lewinnek and Callanan safe zones. The discrepancies of the bilateral combined offset and full leg length were 5.7 and 1.5 mm, which were out of perception in physical activities. These results suggested an improved survival rate and long-term clinical outcome were anticipated in the prospective follow-up. Table 1Changes in key parameter pre- and post-operationParameterPre-opPost-opChangeTK50.1°50.1°Decreased 19.3°LL−11.9°26.4°Changed from kyphosis to lordosis, improved 38.3°PT44.4°25.6°Decreased 18.8°PI64.3°64.3°UnchangedSS19.9°38.7°Increased 18.8°TPA83.5°64.9 Decreased 18.6°SVA41.7 cm10.3 cmDecreased 31.4 cmScoliosis cobb angle81.7°26.6°Corrected 55.1°LL, lumbar lordosis; PI, pelvic incidence; PT, pelvic tilt; SS, sacral slope; SVA, sagittal vertical axis; TK, thoracic kyphosis; TPA, total pelvic angle. Changes in key parameter pre- and post-operation LL, lumbar lordosis; PI, pelvic incidence; PT, pelvic tilt; SS, sacral slope; SVA, sagittal vertical axis; TK, thoracic kyphosis; TPA, total pelvic angle. No incision complications, infection, dislocation, and prosthesis loosening or subsidence were observed during the follow-up. The patient complained that he could not sit upright at 5 months after surgery, and the passive ROM was tested to be 0–60°, which was improved to 0-100° after manual manipulation under anesthesia. Discussion Key findings An AI-assisted preoperative design was applied to address severe hip-spine deformity, which simulated key data (e.g., femoral neck osteotomy location, acetabular cup orientation/size, center of rotation height, combined offset) via CT and 3D reconstruction, reducing complications from poor hip prosthesis position and postoperative soft tissue imbalance (5,6). AI preoperative design could simulate the actual position of the hip prosthesis after pelvic correction post-spine surgery, solving the difficulty of predicting dynamic pelvis-spine changes and facilitating acetabular component implantation into a safe zone (7,8). THA via the supine DAA was conducive to the accurate implementation of the AI-derived plan, restoring hip joint biomechanics and improving motor ability (9). Strengths and limitations Strengths ❖ AI preoperative design provides a relatively accurate surgical plan (7); ❖ AI preoperative design provides a relatively accurate surgical plan (7); ❖ The DAA approach is minimally invasive, aligns with damage-control principles, and promotes rapid recovery (9); ❖ The DAA approach is minimally invasive, aligns with damage-control principles, and promotes rapid recovery (9); ❖ Supine THA accurately restores hip biomechanics and enhances motor function (10,11). ❖ Supine THA accurately restores hip biomechanics and enhances motor function (10,11). Limitations The findings of this case report may not be generalizable to the standard management of hip ankylosis in patients with AS. A more comprehensive assessment of spine-pelvic-hip alignment is warranted to formulate patient-specific surgical strategies. Additionally, extended clinical follow-up is required to strengthen the persuasiveness of the study’s conclusion. Comparison with similar research Past research reported that the AI-HIP system enables comprehensive 3D hip joint evaluation in THA preoperative planning, with higher accuracy and effectiveness than traditional 2D template measurement. Additionally, this AI deep learning system has been applied in preoperative planning for complex hip surgeries such as revision. In contrast, the present study is the first to integrate AI planning and navigation with DAA-THA specifically for managing complex hip-spine syndrome, complemented by a 30-month long-term follow-up to verify clinical outcomes, expanding the application scenario of AI-assisted joint surgery (5). Explanations of findings Severe hip-spine deformity involves impaired hip-spine adaptation and dynamic pelvis-spine changes, which are difficult to predict. AI overcomes this by utilizing CT and 3D reconstruction to simulate key prosthesis parameters, reducing complications related to improper component positioning and soft tissue imbalance (6). For patients with bilateral hip ankylosis at 45° flexion, prone-position spinal corrective surgery cannot be performed in the first stage. AI preoperative design simulates the prosthesis position after pelvic correction post-spine surgery, providing the surgeon with hip component orientation and relative angles, thereby simplifying complex surgery and improving accuracy. Supine-position THA, as supported by existing literature, accurately restores hip joint biomechanics and improves motor ability. The DAA approach, as the most minimally invasive hip surgical approach, offers advantages in damage control and rapid postoperative recovery for complex hip surgeries (10,11). Implications and actions needed This study provides a digitalized and intelligent treatment solution paradigm for complex hip-spine syndrome (especially those secondary to AS), verifying the feasibility and efficacy of combining AI planning, navigation, and minimally invasive DAA-THA. Further accumulation of clinical cases is required to validate the generalizability of the AI-assisted DAA-THA protocol in larger cohorts; promotion of this integrated technique in clinical practice for similar complex hip disorders; continuous optimization of the AI algorithm based on long-term follow-up data to enhance preoperative planning accuracy. Conclusions A staged intervention strategy, integrating AI-driven preoperative planning, computer navigated THA via supine DAA, and subsequent spinal corrective surgery guided by 3D-printed model, is a feasible and effective solution for managing complex hip-spine syndrome secondary to AS under spine-pelvic-hip alignment philosophy. Supplementary The article’s supplementary files as Available at https://aoj.amegroups.com/article/view/10.21037/aoj-2025-1-94/prf