Smartphone-Based Automated Photogrammetry for Reconstruction of Residual Limb Models in Prosthetic Design.

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Prosthetics and Orthotics

Background:

• Accurate 3D modeling of residual limbs is crucial for prosthetic socket design.
• Current scanning methods are often expensive, operator-dependent, and not suitable for frequent clinical use.

Purpose of the Study:

• To develop and validate a low-cost, automated photogrammetry workflow for accurate 3D residual limb modeling.
• To assess the feasibility of using smartphone video for clinical prosthetic applications.

Main Methods:

• A fully automated photogrammetry pipeline using smartphone video or digital camera images.
• Integration of adaptive frame selection, deep learning background removal, ArUco marker scaling, and open-source SfM/MVS reconstruction.
• Validation using 3D-printed phantoms and CT-derived meshes for accuracy and repeatability assessment.

Main Results:

• Smartphone video and full-frame camera acquisitions yielded sub-millimeter surface accuracy.
• Volume and perimeter errors were within ±1%, meeting clinical thresholds for prosthetic socket fabrication.
• High inter-session repeatability was achieved; smartphone still-photo reconstructions showed lower accuracy and stability.

Conclusions:

• Smartphone video-based photogrammetry offers a practical, scalable, and clinically viable method for residual limb modeling.
• This approach is particularly beneficial for resource-constrained or remote healthcare settings.
• The workflow requires no manual post-processing or proprietary software, enhancing accessibility.