Gait laboratory measurement of whole-body kinematics and ground reaction forces during a wide range of activities is frequently performed in joint replacement patient diagnosis, monitoring, and rehabilitation programs. These data are commonly processed in musculoskeletal modeling platforms such as OpenSim and Anybody to estimate muscle and joint reaction forces during activity. However, the processing required to obtain musculoskeletal estimates can be time consuming, requires significant expertise, and thus seriously limits the patient populations studied. Accordingly, the purpose of this study was to evaluate the potential of deep learning methods for estimating muscle and joint reaction forces over time given kinematic data, height, weight, and ground reaction forces for total knee replacement (TKR) patients performing activities of daily living (ADLs). 70 TKR patients were fitted with 32 reflective markers used to define anatomical landmarks for 3D motion capture. Patients were instructed to perform a range of tasks including gait, step-down and sit-to-stand. Gait was performed at a self-selected pace, step down from an 8” step height, and sit-to-stand using a chair height of 17”. Tasks were performed over a force platform while force data was collected at 2000 Hz and a 14 camera motion capture system collected at 100 Hz. The resulting data was processed in OpenSim to estimate joint reaction and muscle forces in the hip and knee using static optimization. The full set of data consisted of 135 instances from 70 patients with 63 sit-to-stands, 15 right-sided step downs, 14 left-sided step downs, and 43 gait sequences. Two classes of neural networks (NNs), a recurrent neural network (RNN) and temporal convolutional neural network (TCN), were trained to predict activity classification from joint angle, ground reaction force, and anthropometrics. The NNs were trained to predict muscle and joint reaction forces over time from the same input metrics. The 135 instances were split into 100 instances for training, 15 for validation, and 20 for testing.Introduction
Methods
The magnitude of principal strain is indicative of the risks of femoral fracture,1,2 while changes in femoral strain energy density (SED) after total hip arthroplasty (THA) have been associated with bone remodeling stimulus.3 Although previous modeling studies have evaluated femoral strains in the intact and implanted femur under walking loads through successfully predicting physiological hip contact force and femoral muscle forces,1,2,3 strains during ‘high load’ activities of daily living have not typically been evaluated. Hence, the objective of this study was to compare femoral strain between the intact and the THA implanted femur under peak loads during simulated walking, stair descent, and stumbling. CTs of three cadaveric specimens were used to develop finite element (FE) models of intact and implanted femurs. Implanted models included a commercially-available femoral stem (DePuy Synthes, Warsaw, IN, USA). Young's moduli of the composite bony materials were interpolated from Hounsfield units using a CT phantom and established relationships.4 Peak hip contact force and femoral muscle forces during walking and stair descent were calculated using a lower extremity musculoskeletal model5 and applied to the femur FE models (Fig. 1). While maintaining the peak hip contact forces, muscle forces were further adjusted using an iterative optimization approach in FE models to reduce the femur deflection to the reported physiological range (< 5 mm).2 Femoral muscle forces during stumbling were estimated utilizing the same optimization approach with literature-reported hip contact forces as input.6 Maximum and minimum principal strains were calculated for each loading scenario. Changes in SED between intact and THA models were calculated in bony elements around the stem.INTRODUCTION
METHODS
Real-time tracking of surgical tools has applications for assessment of surgical skill and OR workflow. Accordingly, efforts have been devoted to the development of low-cost systems that track the location of surgical tools in real-time without significant augmentation to the tools themselves. Deep learning methodologies have recently shown success in a multitude of computer vision tasks, including object detection, and thus show potential for the application of surgical tool tracking. The objective of the current study was to develop and evaluate a deep learning-based computer vision system using a single camera for the detection and pose estimation of multiple surgical tools routinely used in both knee and hip arthroplasty. A computer vision system was developed for the detection and 6-DoF pose estimation of two surgical tools (mallet and broach handle) using only RGB camera frames. The deep learning approach consisted of a single convolutional neural network (CNN) for object detection and semantic key point prediction, as well as an optimization step to place prior known geometries into the local camera coordinate system. Inference on a camera frame with size of 256-by-352 took 0.3 seconds. The object detection component of the system was evaluated on a manually-annotated stream of video frames. The accuracy of the system was evaluated by comparing pose (position and orientation) estimation of a tool with the ground truth pose as determined using three retroreflective markers placed on each tool and a 14 camera motion capture system (Vicon, Centennial CO). Markers placed on the tool were transformed into the local camera coordinate system and compared to estimated location.Introduction
Methods
Alignment of the acetabular cup and femoral components directly affects hip joint loading and potential for impingement and dislocation following total hip arthroplasty (THA) [1]. Changes to the lines of action and moment generating capabilities of the muscles as a result of component position may influence overall patient function. The objectives of this study were to assess the effect of component placement on hip joint contact forces (JCFs) and muscle forces during a high demand step down task and to identify important alignment parameters using a probabilistic approach. Three patients following THA (2 M: 28.3±2.8 BMI; 1 F: 25.7 BMI) performed lower extremity maximum isometric strength tests and a step down task as part of a larger IRB-approved study. Patient-specific musculoskeletal models were created by scaling a model with detailed hip musculature [2] to patient segment dimensions and mass. For each model, muscle maximum isometric strengths were optimized to minimize differences between model-predicted and measured preoperative maximum isometric joint torques at the hip and knee. Baseline simulations used patient-specific models with corresponding measured kinematics and ground reaction forces to predict hip JCFs and muscle forces using static optimization. To assess the combined effects of stem and cup position and orientation, a 1000 trial Monte Carlo simulation was performed with input variability in each degree of freedom based on the ±1 SD range in component placement relative to native geometry reported by Tsai et al. [3] (Figure 1). Maximum confidence bounds (1–99%) were predicted for the hip JCF magnitude and muscle forces for three prime muscles involved in the task (gluteus medius, gluteus minimus and psoas). HJC confidence bounds were compared to Orthoload measurements from telemetric implants from 6 patients performing the step down task. Sensitivity of hip JCF and muscle force outputs was quantified by Pearson Product-Moment correlation between the input parameter and the value of each output averaged across four points in the cycle.Introduction
Methods
