Patelloscope

Finite Element Analysis of Digital Twin MRI to STL Model
Surgical Planning Results
Finite Element Analysis Results

Inspiration

Knee injuries are among the most common musculoskeletal conditions worldwide, affecting athletes, older adults, and everyday people alike. MRI has become excellent at identifying what is damaged in the knee such as ACL tears, meniscal injuries, or cartilage degeneration.

However, after imaging, clinicians still face a critical unanswered question:

How should this specific knee be loaded safely during recovery?

Today, rehabilitation decisions rely on generalized protocols, visual assessment, and clinical experience. These approaches work for many patients, but a significant number experience reinjury, instability, or prolonged pain because internal joint loading cannot be directly observed.

We created Patelloscope, a physics based diagnostic and rehabilitation framework that transforms MRI data into a patient specific digital knee model. By simulating real movements using finite element analysis, Patelloscope helps clinicians understand how injured tissues behave under load and guides safer, personalized rehabilitation strategies.

What it does

Patelloscope is a lower limb diagnostic and rehabilitation support system focused on the knee joint. It creates a patient specific knee digital twin using MRI and CT data and performs finite element analysis to simulate how forces act on the joint during daily activities and rehabilitation exercises.

Patelloscope can: --> Simulate standing, walking, and running gait cycles

--> Analyze stress and strain in ligaments, cartilage, menisci, tendons, and bone

--> Model injured structures such as torn ACL, PCL, MCL, LCL, or meniscal damage

--> Track how stress distributions change as tissues heal over time

--> Recommend safer exercise variants tailored to the individual knee

The goal is not to eliminate stress. Mechanical loading is essential for healing. The goal is to avoid harmful stress concentrations while encouraging safe progression.

How we built it Patient-specific knee digital twin construction

Patelloscope constructs a patient-specific knee digital twin directly from medical imaging, with the goal of preserving mechanical fidelity rather than relying on idealized joint models.

The digital twin is generated from:

MRI-derived joint geometry, including femoral and tibial bone surfaces and cartilage layers, enabling anatomically accurate contact interfaces

Tissue-specific material models parameterized from peer-reviewed biomechanical literature, capturing the relative stiffness of cartilage, bone, and supporting structures

Injury-aware mechanical modifications, such as reduced ligament constraint stiffness, altered joint stability, and asymmetric load transfer to reflect pathological states (e.g., ACL deficiency)

Rather than representing a generic healthy knee, the model explicitly encodes injury-altered joint mechanics, allowing downstream simulations to reflect how force pathways and contact stresses are redistributed in compromised tissue.

Finite element analysis of rehabilitation movements

Rehabilitation exercises are treated as mechanical boundary-value problems, not predefined workouts.

Each movement is parameterized and mapped to boundary conditions applied to the digital twin, including:

Joint flexion angle (e.g., squat depth)

Medial–lateral load distribution (stance width proxy)

Applied compressive and shear loads

Loading rate proxies corresponding to tempo

For each parameterized movement state, quasi-static finite element analysis is performed to solve for internal joint mechanics.

The FEA pipeline computes:

Cartilage–cartilage contact stress distributions, capturing peak pressure and stress gradients across joint compartments

Ligament force and strain proxies, reflecting changes in joint stability under load

Localized stress concentrations within injury-sensitive regions, identifying mechanical risk zones rather than relying on global averages

These simulations allow direct comparison between movement variants by isolating how small changes in mechanics alter internal tissue loading.

Physics-to-decision translation pipeline

Raw finite element outputs are converted into structured, machine-readable biomechanical summaries rather than interpreted heuristically.

Each simulation produces a standardized JSON representation containing:

Regional cartilage stress metrics (peak, mean, high-stress area fraction)

Relative changes in joint constraint behavior due to injury modeling

Longitudinal indicators of mechanical exposure across repeated sessions

This structured biomechanical data is passed to a retrieval-augmented generation (RAG) system, powered by Moorcheh.ai, which grounds interpretation in peer-reviewed rehabilitation and biomechanics literature.

Rather than generating exercises arbitrarily, the system:

Retrieves evidence-based tolerance ranges for specific knee tissues

Maps simulated stress profiles to known safe and progressive loading regimes

Passes grounded context to Gemini, which selects from a finite, clinically accepted exercise set

The output is a structured recommendation that explains why certain movement variants are mechanically safer or riskier for a given patient state, rather than prescribing exercises blindly.

How clinicians benefit

Patelloscope provides mechanical insight that static imaging cannot offer.

By integrating finite element analysis with patient-specific anatomy, clinicians can:

Visualize how injured ligaments and cartilage redistribute load under rehabilitation exercises

Quantitatively assess whether proposed movements exceed tissue-specific stress tolerances

Compare multiple movement variants before prescribing them, rather than relying on trial-and-error

Understand how alignment, anatomy, and injury severity influence internal joint mechanics

Track mechanical recovery trends over time, instead of relying solely on pain or subjective feedback

This enables a shift from symptom-driven rehabilitation to mechanics-informed decision support, reducing reinjury risk and improving personalization, particularly for older adults or patients with heterogeneous bone density and tissue quality.

Exercise recommendation pipeline Simulation outputs are converted into structured JSON containing: --> Regional stress values

--> Tissue stiffness changes

--> Healing progression indicators

This data is sent to a RAG database powered by Moorcheh.ai. By sourcing information from peer-reviewed rehabilitation literature, it outputs key information regarding optimal stress loads on the different knee structures as well as methods of recovery. It passes this information to Gemini which outputs a list of optimal exercises alongside clinical reasoning in a structured format. The system selects from a finite, evidence-based exercise set and explains why certain variants are safer.

How clinicians benefit

Patelloscope provides clinicians with insight that imaging alone cannot offer. With finite element analysis, clinicians can: --> Visualize how damaged ligaments behave under load

--> Evaluate whether rehab exercises exceed tissue tolerance

--> Compare movement variants before prescribing them

--> Quantify how anatomy and alignment influence injury mechanics

--> Track mechanical recovery over time rather than relying solely on pain

This supports personalized medicine and reduces the risk of reinjury, especially in older adults or patients with variable bone density and tissue quality.

Challenges we ran into

--> Translating complex biomechanical concepts into interpretable outputs

--> Simplifying tissue material models without losing relative accuracy

--> Designing simulations that are computationally feasible yet clinically meaningful

--> Bridging engineering outputs with clinician friendly explanations

We learned that finite element analysis is powerful, but only useful if it remains understandable and actionable.

Accomplishments that we're proud of

--> Designing a full injury aware knee digital twin framework

--> Integrating biomechanics, imaging, and rehabilitation into one system

--> Creating a physics driven recommendation engine grounded in scientific literature

--> Demonstrating how FEA can be applied beyond research labs and into clinical decision support

What we learned

--> MRI answers what is damaged, not how the knee should move

--> Rehabilitation is fundamentally a mechanical problem

--> Personalized recovery requires understanding internal joint loading

--> Physics based modeling can meaningfully augment clinical judgment

What's next for Patelloscope

--> Expanding to hip and ankle joints

--> Incorporating real time gait data from wearable sensors

--> Validating simulations against motion capture and force plate data

--> Clinical pilot studies with physiotherapy clinics

--> Longitudinal modeling of tissue remodeling and stiffness recovery

Patelloscope represents a step toward physics based digital twins for rehabilitation, enabling safer, more personalized recovery for millions of people with knee injuries.