Detecting total hip replacement prosthesis design on plain radiographs using deep convolutional neural network

Abstract

Identifying the design of a failed implant is a key step in the preoperative planning of revision total joint arthroplasty. Manual identification of the implant design from radiographic images is time‐consuming and prone to error. Failure to identify the implant design preoperatively can lead to increased operating room time, more complex surgery, increased blood loss, increased bone loss, increased recovery time, and overall increased healthcare costs. In this study, we present a novel, fully automatic and interpretable approach to identify the design of total hip replacement (THR) implants from plain radiographs using deep convolutional neural network (CNN). CNN achieved 100% accuracy in the identification of three commonly used THR implant designs. Such CNN can be used to automatically identify the design of a failed THR implant preoperatively in just a few seconds, saving time and improving the identification accuracy. This can potentially improve patient outcomes, free practitioners' time, and reduce healthcare costs.

Comparing performance of deep convolutional neural network with orthopaedic surgeons on identification of total hip prosthesis design from plain radiographs

“…Other studies have applied DL methods on plain film radiographs for various orthopedic applications [7][8][9][10][11][12][13][14][15][16] . However, to the best of our knowledge, our pilot study identifying three types of THR femoral implant designs 18 , and this study identifying nine types of THR femoral implant designs are the first applications of DL method to automatically detect THR femoral implant designs. We demonstrated that the CNN generally achieved on-par accuracy with the surgeons in identifying these THR implant designs.…”

“…Online data augmentation created realistic new x-rays, preserving the same constant design of each THR implant, while compensating for intrinsic x-rays variations. We successfully followed this online data augmentation method in our previous work 18 with smaller range (horizontal flip, rotation between 0° and 25°, width and height shift between 0% and 15%, shearing between 0% and 10%, and magnification between 0% and 15%), and less number of epochs (350 epochs). We had to increase both the data augmentation range and number of epochs in this study due to the more complex classification task compared to our previous study (nine THR implant designs in this study vs. three THR implant designs in our previous study).…”

“…In recent years, DL methods have been applied to the interpretation of plain film radiographs with high degrees of success for identification and classification of orthopaedic fractures, staging knee osteoarthritis (OA) severity, and detection of aseptic loosening, to name a few [7][8][9][10][11][12][13][14][15][16][17] . In a previous pilot study, we successfully trained a DL method for the first time to classify a given THR radiograph into one of three possible femoral component designs 18 . Although, this represented a limited task and we did not compare the model's performance with orthopaedic surgeons, the results were highly encouraging with the model achieving 100% accuracy.…”

Deep Learning in Orthopedics: How Do We Build Trust in the Machine?

“…2 Machine learning itself can be further divided into more classical algorithms (e.g., support vector machine, decision tree, and neutral network) to extract knowledge from tabulated data sets, [2][3][4][5] and more recently developed ''deep learning'' algorithms (e.g., convolutional neural network [CNN]) to extract knowledge from imaging data sets. 6 Since orthopedic diagnosis and prognosis rely heavily on manual interpretation of medical images (X-ray, computed tomography scans, and magnetic resonance imaging), the application of AI in orthopedics has mainly focused on implementation of deep learning on these images. Deep learning can help radiologists and orthopedic surgeons with automatic interpretation of medical images that can potentially improve the diagnostic accuracy and speed, flag the most critical and urgent patients for immediate attention, reduce the amount of human error due to fatigue and/or inexperience, reduce the strain on medical professionals by reducing their workload, and, in general, improve orthopedic care.…”

“…Deep learning has been successfully used in different orthopedic applications such as fracture detection, [7][8][9][10] bone tumor diagnosis, 11 detecting hip implant mechanical loosening, 6 and grading osteoarthritis. 12,13 The implementation of deep learning in orthopedics is not without challenges and limitations.…”