Robust automatic hexahedral cartilage meshing framework enables population-based computational studies of the knee

Abstract

Osteoarthritis of the knee is increasingly prevalent as our population ages, representing an increasing financial burden, and severely impacting quality of life. The invasiveness of in vivo procedures and the high cost of cadaveric studies has left computational tools uniquely suited to study knee biomechanics. Developments in deep learning have great potential for efficiently generating large-scale datasets to enable researchers to perform population-sized investigations, but the time and effort associated with producing robust hexahedral meshes has been a limiting factor in expanding finite element studies to encompass a population. Here we developed a fully automated pipeline capable of taking magnetic resonance knee images and producing a working finite element simulation. We trained an encoder-decoder convolutional neural network to perform semantic image segmentation on the Imorphics dataset provided through the Osteoarthritis Initiative. The Imorphics dataset contained 176 image sequences with varying levels of cartilage degradation. Starting from an open-source swept-extrusion meshing algorithm, we further developed this algorithm until it could produce high quality meshes for every sequence and we applied a template-mapping procedure to automatically place soft-tissue attachment points. The meshing algorithm produced simulation-ready meshes for all 176 sequences, regardless of the use of provided (manually reconstructed) or predicted (automatically generated) segmentation labels. The average time to mesh all bones and cartilage tissues was less than 2 min per knee on an AMD Ryzen 5600X processor, using a parallel pool of three workers for bone meshing, followed by a pool of four workers meshing the four cartilage tissues. Of the 176 sequences with provided segmentation labels, 86% of the resulting meshes completed a simulated flexion-extension activity. We used a reserved testing dataset of 28 sequences unseen during network training to produce simulations derived from predicted labels. We compared tibiofemoral contact mechanics between manual and automated reconstructions for the 24 pairs of successful finite element simulations from this set, resulting in mean root-mean-squared differences under 20% of their respective min-max norms. In combination with further advancements in deep learning, this framework represents a feasible pipeline to produce population sized finite element studies of the natural knee from subject-specific models.

An Automated and Robust Tool for Musculoskeletal and Finite Element Modeling of the Knee Joint

“…The hexahedral meshing algorithm and FE model: We adopted a previously developed hexahedral meshing algorithm [22], [29] to create a structured mesh for the femoral, tibial, and patellar cartilages and menisci from the auto-segmented geometries (Fig. 1-D).…”

“…1-D). The algorithm used a swept set of point origins and raytracing techniques to construct rectangular grids [22], [29]. However, we slightly modified the sweeping functions (without altering the total mesh volume) to mesh the cartilages and menisci according to the tissue structure and composition while considering the FE modeling bottlenecks and computation costs.…”

“…However, these methods are not fully automated, e.g., they require manual measurement of knee anatomical dimensions from medical images. Fully automated knee FE modeling algorithms have been developed based on auto-segmentation of magnetic resonance imaging (MRI) of subjects' knees [22], [23]. Nonetheless, these offer no potential to produce the subject's MSK model [22], [23], are limited to a particular MRI sequence [22], [23], exclude some knee tissues such as menisci [22], can take hours to create the subject's FE model [23], and more importantly, often require manual preprocessing of the medical images [23].…”

“…Alterations in inputs to the FE model (e.g., obtained from different MSK modeling frameworks) [14] and excluding tissues such as the menisci [24] can substantially alter the FE model's estimated knee biomechanics. To the authors' knowledge, the existing automatic knee modeling approaches use rigid or linear elastic material models for knee cartilage and menisci [22], [23], which can largely affect the estimated knee biomechanics [25]. Notably, a nonlinear composite material model is essential to simultaneously assess mechanical responses of both fibrillar (collagen) and nonfibrillar (proteoglycans) matrices within the cartilage and menisci [6], [25], [26].…”

“…To improve the robustness of the tool, we developed two modeling pipelines, consisting of 1) template-based and 2) auto-meshing. To this end, we leveraged previously developed and validated template-based [17], [21] as well as hexahedral meshing [22], [29] algorithms. We compared knee biomechanics estimated by the two MSK and FE modeling pipelines and two different MRI sequences, as well as against a manually created FE model.…”