Preoperative Planning Software in Orthopaedic Surgery: A 2026 Guide
A practical guide to modern preoperative planning software for orthopaedic surgery — from 2D templating to AI-powered 3D reconstruction, and what features matter for clinical workflow.
From Templates to Intelligence
Orthopaedic preoperative planning has evolved through three distinct generations. The first — acetate template overlay on printed radiographs — served surgeons for decades but offered no patient-specific adaptation. The second brought templating to digital platforms, enabling on-screen measurement and implant sizing from DICOM images. The third generation, now emerging, integrates AI to automate segmentation, landmark detection, angle measurement, and outcome prediction from standard clinical imaging.
This evolution matters because preoperative planning directly impacts surgical precision. A 2023 study demonstrated that 3D preoperative planning can improve orthopaedic surgery precision by up to 44.7%. AI-based planning has been shown to reduce manual plan corrections by approximately 40% compared to manufacturer standard plans.
The question for surgeons in 2026 is no longer whether to use digital planning tools, but which generation of technology serves their clinical needs — and whether the AI capabilities now available justify the transition.
The Competitive Landscape
Several platforms compete for the orthopaedic preoperative planning market, each with different architectural approaches and clinical focus areas.
Ortoma (Sweden) offers CE-marked 3D planning for total hip and knee arthroplasty. Their platform uses cloud-based CT segmentation and provides surgeon-adjustable implant positioning with range-of-motion simulation. Ortoma's strength is their regulatory clearance and clinical validation in arthroplasty planning.
Enhatch (USA) positions itself as an AI-native surgical planning platform, automating segmentation and implant templating across multiple joint applications. Their cloud architecture enables powerful server-side processing but requires image upload to external infrastructure.
Proprio (USA) focuses on intraoperative navigation rather than preoperative planning, using light-field technology to provide real-time 3D visualisation during surgery. While not a direct planning competitor, their approach represents the convergence of preoperative and intraoperative intelligence.
TraumaCad / Brainlab represents the established enterprise segment — comprehensive planning suites integrated into hospital PACS infrastructure. These platforms offer broad functionality but typically require institutional procurement, IT integration, and significant licensing investment.
The emerging alternative is browser-based planning that runs entirely client-side — no installation, no image upload, no server-side patient data. This approach sacrifices the computational headroom of cloud platforms but eliminates data governance complexity entirely.
What Modern Preoperative Planning Software Should Do
The Imaging Foundation: DICOM-Native Processing
The foundation of any planning platform is its ability to work directly with DICOM files. Software that requires JPEG or PNG export loses the metadata that makes calibrated measurements possible — pixel spacing, slice thickness, patient positioning, and acquisition parameters.
A DICOM-native platform reads this metadata automatically, ensuring that a joint space width measurement is physically calibrated rather than pixel-estimated.
Automated Segmentation: Hours to Minutes
Manual bone segmentation — tracing the contours of femur, tibia, and patella across dozens or hundreds of CT slices — is the most time-consuming step in traditional 3D planning workflows. An experienced technician may spend 2–4 hours on a single knee segmentation.
Deep learning models (nnU-Net, SAM-Med3D, TotalSegmentator) have reduced this to minutes. These models learn to recognise bone boundaries from thousands of annotated CT volumes and can segment complex anatomy with accuracy metrics approaching expert manual segmentation. The clinical impact is not just time savings — it makes 3D planning economically viable for routine cases, not just complex revisions.
The quality question is whether the segmentation is editable. Automated segmentation will occasionally miss anatomy or include artefacts. The surgeon or technician must be able to manually correct the output — a workflow that combines AI speed with human verification.
Geometric Analysis: Angles, Axes, and Alignment
Preoperative assessment for procedures like high tibial osteotomy (HTO) or distal femoral osteotomy (DFO) requires precise measurement of mechanical and anatomical angles. The key measurements include mechanical axis deviation (HKA), medial proximal tibial angle (MPTA), lateral distal femoral angle (LDFA), posterior tibial slope, and femoral-tibial angle (FTA).
Manual measurement of these angles is subjective and time-consuming. AI-powered landmark detection can identify anatomical reference points — femoral head centre, knee centre, ankle centre — and compute angles automatically. This does not eliminate the need for surgical judgment in interpreting the measurements, but it provides consistent, reproducible baselines that reduce inter-observer variability.
Implant Templating and Size Prediction
Digital templating has evolved from simple overlay to predictive sizing. AI models trained on large implant databases can analyse patient anatomy and predict optimal implant size, reducing intraoperative surprises. The practical impact extends to supply chain: predictive sizing enables hospitals to prepare the predicted sizes and one size above and below to the hospital, rather than a full range. For hospitals managing inventory costs, this is a meaningful operational improvement.
Patient-Specific Instrumentation (PSI)
The integration of AI-powered segmentation with CAD design tools enables the creation of patient-specific surgical guides — custom 3D-printed instrumentation that matches the patient's exact anatomy. PSI guides define cut planes, drill trajectories, and implant positioning based on the preoperative plan.
Our published research in The Orthopaedic Journal of Sports Medicine (OJSM) validated the clinical accuracy of PSI guides for knee reconstruction, demonstrating improved intraoperative precision compared to conventional freehand techniques. The key finding was consistency — PSI reduced the variability in alignment outcomes across different surgeon experience levels.
The Architecture Question: Cloud vs. Local vs. Hybrid
Where preoperative planning software runs has direct implications for performance, privacy, and accessibility.
Server-side platforms upload patient imaging to cloud servers for processing. This enables heavy computational tasks (full CT segmentation, 3D mesh generation) without requiring powerful local hardware. The trade-off: patient data is transmitted to and stored on external servers, creating data processing obligations under KVKK, GDPR, and HIPAA.
Local installations keep all data on-premises but require dedicated hardware, IT support, and software maintenance. This is the traditional enterprise model — maximum data control at maximum operational cost.
Browser-based platforms represent a third approach. The application loads in the surgeon's browser, processing happens locally, and no patient data leaves their device. This provides the accessibility of cloud platforms with the privacy guarantees of local installations.
The Salnus Surgeon Portal follows this browser-based architecture. DICOM files are parsed, rendered, and analysed entirely client-side using Cornerstone3D for GPU-accelerated multiplanar reconstruction and ONNX Runtime Web for AI inference.
AI in Preoperative Planning: What Works and What Doesn't
What Works (Clinical Evidence Supports)
Automated KL grading from weight-bearing knee radiographs. Multiple CE-marked tools exist. Clinical validation shows AI matches or approaches expert radiologist agreement. Our pipeline achieves 84.1% binary accuracy and 70.3% five-class accuracy on an independent test set.
Bone segmentation from CT. Deep learning segmentation of femur, tibia, and patella is reliable enough for clinical planning workflows, with accuracy metrics approaching manual expert segmentation. Tools like TotalSegmentator have been validated across multiple anatomical regions.
Geometric measurement automation. AI-powered landmark detection for mechanical axis and alignment angles reduces measurement time and inter-observer variability. Clinical validation exists for HKA, MPTA, and LDFA measurements.
What's Emerging (Promising but Early)
OA screening from radiographs. Binary OA detection (84–95% accuracy) is clinically useful. Five-class KL grading (65–85% accuracy) is informative but not yet reliable enough to replace expert assessment.
MRI-based ligament assessment. ACL tear detection has achieved research-level accuracy, and PCL assessment is advancing. External validation remains limited.
Outcome prediction. Return-to-sport prediction, revision risk models exist but cross-institutional generalisability is unproven.
What Doesn't Work (Yet)
Fully autonomous surgical planning. No AI system can replace the surgeon's clinical judgment in interpreting imaging findings, weighing patient-specific factors, and formulating a surgical strategy. AI tools that present themselves as replacement rather than augmentation should be approached with caution.
Single-model universality. A model trained on knee OA radiographs does not transfer to hip OA, ACL tears, or spinal pathology. Each clinical application requires dedicated training data, validation, and (for clinical use) regulatory clearance.
Evaluating Preoperative Planning Software: A Checklist
For surgeons evaluating planning software in 2026, these criteria distinguish clinical tools from research demos:
DICOM-native? Does it work directly with DICOM files, preserving calibration metadata?
AI-assisted? Does it automate segmentation, measurement, or grading — and are the results editable?
Explainable? Can you see what the AI is doing — heatmaps, landmark overlays, confidence scores?
Privacy-compliant? Where does patient data go? Client-side processing eliminates the most complex privacy challenges.
Regulatory status? Is it CE-marked for clinical use, or Research Use Only?
Integrated? Does it fit into your existing workflow, or require a separate application and manual data transfer?
Interoperable? Can outputs be exported in standard formats (PDF, STL, DICOM SR) for use in other clinical systems?
The Path Forward
Preoperative planning is evolving from a manual, template-based process to an AI-augmented, patient-specific workflow. The technology is mature enough for clinical adoption in specific applications while remaining nascent in others.
For surgeons, the practical approach is to adopt validated tools for established tasks — digital templating, automated measurement, CT segmentation for complex cases — and to evaluate emerging capabilities through pilot programmes and research collaborations rather than wholesale platform commitments.
The platforms that succeed will be those that augment surgical judgment rather than attempting to replace it, that are transparent about their capabilities and limitations, and that respect patient privacy as a foundational architectural principle rather than a compliance checkbox.
Explore Salnus Surgeon Portal →
Disclaimer: This article is for educational and research purposes only. Salnus tools are designated for Research Use Only (RUO) and are not cleared medical devices. Clinical decisions should be made by qualified physicians based on comprehensive patient assessment. Mention of third-party products is for educational context only.
References:
- Zhou X, et al. 3D preoperative planning improves precision of orthopaedic surgeries by up to 44.7%. Int J Surg. 2023.
- Lambrechts A, et al. AI-based preoperative plans reduce manual corrections by 39.71%. J Arthroplasty. 2022.
- Salnus Research Group. 3D-Printed Patient-Specific Guides for Knee Reconstruction. OJSM. 2026.
- Elkohail A, et al. AI-Enhanced Surgical Decision-Making in Orthopedics. Cureus. 2025.
Reviewed by the Salnus biomedical engineering team.