Mobile Assisted Reconstruction In Orthopaedics (M.A.R.I.O):expanding to X-ray based navigation

Research Question Can we adapt the successful M.A.R.I.O surgical navigation system from 3D CT-based to 2D X-ray-based for accurate, even lower-cost surgical navigation in total knee arthroplasty (TKA)? Background TKA is a common surgical procedure, with over 103,000 cases in the UK and 1.3 million in the US in 2022. Despite its prevalence, revision rates remain high (7% at 5 years; 12% at 10 years), and 20% of patients report dissatisfaction. Poor implant alignment is a key contributing factor. While robotic systems improve alignment, their cost and complexity limit widespread use, especially in lower-resource hospitals. Over 80% of TKAs globally still use conventional X-ray-based planning, which lacks 3D precision. Smart Surgical Solutions (S3), a spinout from ICL, has developed M.A.R.I.O — a low-cost, smartphone-based navigation device that uses CT scans for high-precision alignment. However, reliance on CT limits scalability. This project aims to develop an AI model to reconstruct 3D bone geometry from 2D X-rays and integrate it into M.A.R.I.O, enabling scalable and affordable 3D planning without CT. Aims and Objectives This project aims to transform the M.A.R.I.O system into a CT-free solution for surgical planning and navigation. The project has two main objectives: 1.AI Model Development: to reconstruct 3D bone geometry from X-rays, and integrate it into system, enabling X-ray-based real-time M.A.R.I.O surgical navigation. 2.System Performance Validation: through surrogate bone and cadaveric experiments, to examine its precision and reliability. Methods The project is structured into four interlinked work packages over 12 months: WP1(M1–6): AI Model Development – Synthetic X-ray datasets (n>2000) will be generated from our available CT scans for model training, using data augmentation for robustness. Accuracy will be evaluated using Hausdorff Distance ( 98%). The AI model will be integrated into M.A.R.I.O to support pre-surgical planning and real-time navigation. WP2 (M5-12): Performance Evaluation – A structured cadaver study protocol will be established, comparing X-ray-based and current CT-based navigation results. Usability testing with >5 surgeons will inform further suggestions on iterative refinement. WP3 (M1-12): Regulatory Strategy & Project Management – Regulatory compliance under ISO 13485 and EU AI Act will be maintained. We will also develop a clinical trial plan, ensure NHS data compatibility. WP4 (M1-12): PPIE–The development of this project is guided by our ongoing PPIE efforts, and we are committed to maintaining this engagement throughout the entire lifecycle to ensure that M.A.R.I.O is aligned with patient values, needs, and expectations. Knowledge Mobilisation, Dissemination, and Impact Findings will be shared through peer-reviewed publications, international conferences, and local community engagement. We will involve NHS staff and PPIE groups to ensure alignment with clinical workflows and patient needs. Successful project completion will enable clinical trial funding through i4i PDA and private investment, advancing M.A.R.I.O toward UKCA certification and NHS adoption. Research Inclusion Our image dataset includes diverse populations (UK, East Asian, US & EU) to support generalisable AI model development. Surgeon and patient voices with different background will be incorporated throughout the project to ensure inclusivity and usability across NHS settings.