In a groundbreaking convergence of biomechanics and artificial intelligence, researchers from the New Jersey Institute of Technology (NJIT), in collaboration with the Veterans Affairs New York Harbor Healthcare System and Rutgers New Jersey Medical School, are pioneering novel methodologies to significantly improve the fit and comfort of prosthetic lower limbs. This interdisciplinary effort addresses a critical and persistent challenge faced by amputees worldwide, particularly veterans, who often suffer from debilitating skin complications resulting from imperfect prosthetic interfaces.
For individuals reliant on lower-limb prostheses, the interface between the residual limb and the prosthetic socket is a focal point of potential irritation and injury. Despite advancements in prosthetic liner technology designed to cushion and stabilize the interaction, approximately three-quarters of these patients experience adverse skin conditions such as blisters, calluses, irritations, ulcers, or open wounds at the limb-socket junction. These complications not only diminish quality of life but also obstruct the ability to use the prosthesis effectively, leading to reduced mobility and secondary health issues.
Dr. Jason Maikos, a biomechanical engineer with the Veterans Affairs, underscores the inherent difficulty of achieving a perfect socket fit. The residual limb, due to its anatomical complexity and variability in soft tissue characteristics, inevitably moves within the socket during ambulation. This movement involves multi-directional translations and rotations that, if excessive, generate friction and shear forces on the skin, precipitating tissue damage. The issue is compounded in patient populations with comorbidities such as diabetes, where tissue healing processes are impaired, further escalating the risk of chronic wounds and infections.
To better understand and quantify these biomechanical interactions, researchers employ a sophisticated imaging modality known as dynamic stereo X-ray (DSX). Unlike traditional X-ray imaging, DSX captures synchronized biplanar views, enabling precise three-dimensional tracking of internal bony and soft tissue movement with high temporal and spatial resolution. The technique involves affixing radiopaque markers infused with barium onto the patient’s residual limb, which serves as fiducial points tracked across sequential frames during dynamic activities like walking.
However, the wealth of data generated by DSX presents a formidable analytical challenge. Each imaging session produces a massive volume of images, typically around 1,000 per camera perspective, with dozens of markers whose positions must be meticulously annotated frame by frame. Manually processing this data requires extensive human labor, consuming approximately an entire day per session, which hampers the scalability of this promising diagnostic tool for widespread clinical use.
This bottleneck catalyzed the involvement of NJIT, where Assistant Professor Salam Daher and her research team are harnessing cutting-edge artificial intelligence techniques to automate and accelerate the interpretation of DSX datasets. By training AI models on labeled examples, the system learns to identify and track the radiopaque markers, significantly reducing the analysis time from nearly one full day to approximately fifteen minutes per trial. This leap in efficiency not only expedites research cycles but also opens the possibility of integrating DSX into routine clinical workflows for prosthetic fitting optimization.
Daher’s approach involves evaluating both conventional computer vision algorithms and specialized AI frameworks to determine the most robust and reliable method of marker detection and tracking. Although conventional computer vision has demonstrated some efficacy, AI-driven models hold the promise of improved accuracy and adaptability once sufficiently trained with diverse datasets. The research is ongoing, as neither approach currently guarantees complete reliability, necessitating further exploration of hybrid techniques that combine the strengths of both.
One of the primary technical hurdles encountered is the challenge of depth perception inherent in stereo imaging of dynamic human motion. Accurately interpreting the relative three-dimensional spatial positions of markers on moving limbs, especially when markers from bilateral limbs come into close proximity or occlude one another within the imaging field, requires sophisticated computational models. Fine-tuning AI architectures to surmount these challenges is a focal point of the research.
Further complicating algorithm development is the question of data sufficiency for model training. AI systems often demand large, annotated datasets to generalize effectively to new cases, yet the collection and preparation of such comprehensive DSX imaging data remain resource-intensive. Establishing an optimal balance between data volume and model performance is critical to ensure both feasibility and clinical reliability.
In parallel with these technological innovations, the team recognizes the necessity of delivering results in an accessible and interpretable manner for healthcare providers. Prosthetists, clinicians, and rehabilitation specialists involved in prosthetic care are not necessarily experts in AI or computer vision. Therefore, Daher’s group is designing custom user interfaces that translate complex biomechanical data and AI-derived insights into practical guidance that can inform socket adjustments and personalized prosthetic design.
The project’s collaborative framework is underscored by a $40,000 grant awarded by the American Orthopaedic Foot & Ankle Society, with Dr. David Paglia of Rutgers Medical School serving as principal investigator. Paglia and his colleagues have developed baseline error metrics for the DSX system and actively contribute to interpreting the evolving data and advancing toward clinical adoption. Their combined efforts represent a significant step in bridging experimental imaging techniques with real-world applications that enhance patient outcomes.
Looking ahead, this integration of AI and dynamic imaging may revolutionize prosthetic limb design and fitting processes by enabling real-time, patient-specific assessments of biomechanical interactions. The potential to minimize skin injury risk and enhance comfort could dramatically improve mobility, independence, and quality of life for millions living with limb loss. Moreover, this methodological synergy offers a compelling blueprint for leveraging AI in other domains of biomechanical health and rehabilitation medicine.
As the project advances, continued interdisciplinary collaboration will be essential to navigate technical challenges, validate AI models across diverse patient populations, and ultimately translate research innovations into practical devices and protocols. The promise of real-time, AI-enhanced prosthetic fitting heralds a new era where technology profoundly empowers human resilience and mobility.
Subject of Research: Artificial Intelligence-Enabled Analysis of Dynamic Stereo X-Ray Imaging for Improving Prosthetic Fit and Comfort in Lower Limb Amputees
Article Title: Revolutionizing Prosthetic Limb Care: AI Accelerates Dynamic Stereo X-Ray Analysis to Enhance Socket Fit and Patient Outcomes
News Publication Date: Not specified
Web References:
https://mediasvc.eurekalert.org/Api/v1/Multimedia/1a255ca9-3d7a-44e0-ba7c-4f5e3f62a170/Rendition/low-res/Content/Public
Image Credits: U.S. Consumer Product Safety Commission
Keywords
Prosthetics, Dynamic Stereo X-Ray (DSX), Artificial Intelligence, Biomechanics, Lower Limb Amputation, Socket Fit, Medical Imaging, Computer Vision, Rehabilitation, Veterans Health, Diabetes Complications, Prosthetic Comfort
