Knowledge-guided artificial intelligence supports radiographic surgical planning for acetabular reconstruction in developmental hip dysplasia
Acetabular planning for developmental dysplasia of the hip (DDH) commonly uses two-dimensional templating or CT-based workflows. We developed Knowledge-base-guided Acetabular Planning for Developmental Dysplasia of the Hip (KAP-DDH), which integrates automated two-dimensional radiographic analysis with paired historical surgical reconstructions to generate patient-specific plans. In this retrospective two-centre study, development, internal-validation and external-validation cohorts comprised 2,760, 691 and 422 records. Evaluation used two expert references and two blinded surgeon planners. External cup-diameter mean absolute errors were 3.18 mm for KAP-DDH, 4.37 mm for Surgeon 1 and 3.25 mm for Surgeon 2; both surgeons had smaller errors internally. Historical-case guidance reduced cup-diameter error by 0.54 mm internally and 0.62 mm externally versus Crowe-specific defaults and reduced positional errors, with little change in inclination or coverage errors. All evaluated KAP-DDH plans met prespecified inclination and coverage criteria. These findings support translating historical reconstruction experience into patient-specific radiographic planning targets.
Authors
- Shihao Chen (ORCID: https://orcid.org/0000-0002-9123-0375)
- Tao Deng (ORCID: https://orcid.org/0000-0002-8303-0891)
- Yong Nie (ORCID: https://orcid.org/0000-0002-6132-379X)
- Yi Zeng (ORCID: https://orcid.org/0009-0004-1998-2558)
- Junqing Wang
- Jiaqu Sun
- Zongke Zhou (ORCID: https://orcid.org/0000-0001-5049-4801)
- Qiang Zhang
- Guo Chen
- Liang Zhao
- Biao Wang
- Kang Li
- Bingxing Huo
Institutions
- West China Hospital of Sichuan University (CN)
Publication Details
- Journal
- Research Square
- Published
- 2026-11-18
- DOI
- https://doi.org/10.21203/rs.3.rs-11591828/v1
- Primary Topic
- Hip disorders and treatments
- Type
- preprint