Application of deep learning for automated diagnosis and classification of hip dysplasia on plain radiographs.

BACKGROUND: Hip dysplasia is a condition where the acetabulum is too shallow to support the femoral head and is commonly considered a risk factor for hip osteoarthritis. The objective of this study was to develop a deep learning model to diagnose hip dysplasia from plain radiographs and classify dysplastic hips based on their severity. METHODS: We collected pelvic radiographs of 571 patients from two single-center cohorts and one multicenter cohort. The radiographs were split in half to create hip radiographs (n = 1022). One orthopaedic surgeon and one resident assessed the radiographs for hip dysplasia on either side. We used the center edge (CE) angle as the primary diagnostic criteria. Hips with a CE angle < 20°, 20° to 25°, and > 25° were labeled as dysplastic, borderline, and normal, respectively. The dysplastic hips were also classified with both Crowe and Hartofilakidis classification of dysplasia. The dataset was divided into train, validation, and test subsets using 80:10:10 split-ratio that were used to train two deep learning models to classify images into normal, borderline and (1) Crowe grade 1-4 or (2) Hartofilakidis grade 1-3. A pre-trained on Imagenet VGG16 convolutional neural network (CNN) was utilized by performing layer-wise fine-turning. RESULTS: Both models struggled with distinguishing between normal and borderline hips. However, achieved high accuracy (Model 1: 92.2% and Model 2: 83.3%) in distinguishing between normal/borderline vs. dysplastic hips. The overall accuracy of Model 1 was 68% and for Model 2 73.5%. Most misclassifications for the Crowe and Hartofilakidis classifications were +/- 1 class from the correct class. CONCLUSIONS: This pilot study shows promising results that a deep learning model distinguish between normal and dysplastic hips with high accuracy. Future research and external validation are warranted regarding the ability of deep learning models to perform complex tasks such as identifying and classifying disorders using plain radiographs. LEVEL OF EVIDENCE: Diagnostic level IV.

The use of deep learning enables high diagnostic accuracy in detecting syndesmotic instability on weight-bearing CT scanning.

PURPOSE: Delayed diagnosis of syndesmosis instability can lead to significant morbidity and accelerated arthritic change in the ankle joint. Weight-bearing computed tomography (WBCT) has shown promising potential for early and reliable detection of isolated syndesmotic instability using 3D volumetric measurements. While these measurements have been reported to be highly accurate, they are also experience-dependent, time-consuming, and need a particular 3D measurement software tool that leads the clinicians to still show more interest in the conventional diagnostic methods for syndesmotic instability. The purpose of this study was to increase accuracy, accelerate analysis time, and reduce interobserver bias by automating 3D volume assessment of syndesmosis anatomy using WBCT scans. METHODS: A retrospective study was conducted using previously collected WBCT scans of patients with unilateral syndesmotic instability. One-hundred and forty-four bilateral ankle WBCT scans were evaluated (48 unstable, 96 control). We developed three deep learning models for analyzing WBCT scans to recognize syndesmosis instability. These three models included two state-of-the-art models (Model 1-3D Convolutional Neural Network [CNN], and Model 2-CNN with long short-term memory [LSTM]), and a new model (Model 3-differential CNN LSTM) that we introduced in this study. RESULTS: Model 1 failed to analyze the WBCT scans (F1 score = 0). Model 2 only misclassified two cases (F1 score = 0.80). Model 3 outperformed Model 2 and achieved a nearly perfect performance, misclassifying only one case (F1 score = 0.91) in the control group as unstable while being faster than Model 2. CONCLUSIONS: In this study, a deep learning model for 3D WBCT syndesmosis assessment was developed that achieved very high accuracy and accelerated analytics. This deep learning model shows promise for use by clinicians to improve diagnostic accuracy, reduce measurement bias, and save both time and expenditure for the healthcare system. LEVEL OF EVIDENCE: II.

Defining and predicting radiographic knee osteoarthritis progression: a systematic review of findings from the osteoarthritis initiative.

PURPOSE: The purposes of this systematic review were to (1) identify the commonly used definitions of radiographic KOA progression, (2) summarize the important associative risk factors for disease progression based on findings from the OAI study and (3) summarize findings from radiographic KOA progression prediction modeling studies regarding the characterization of progression and outcomes. METHODS: A systematic review was performed by conducting a literature search of definitions, risk factors and predictive models for radiographic KOA progression that utilized data from the OAI database. Radiographic progression was further characterized into "accelerated KOA" and "typical progression," as defined by included studies. RESULTS: Of 314 studies identified, 41 studies were included in the present review. Twenty-eight (28) studies analyzed risk factors associated with KOA progression, and 13 studies created or validated prediction models or risk calculators for progression. Kellgren-Lawrence (KL) grade based on radiographs was most commonly used to characterize KOA progression (50%), followed by joint space width (JSW) narrowing (32%) generally over 48 months. Risk factors with the highest odds ratios (OR) for progression included periarticular bone mineral density (OR 10.40), any knee injury within 1 year (OR 9.22) and baseline bone mineral lesions (OR 7.92). Nine prediction modeling studies utilized both clinical and structural risk factors to inform their models, and combined models outperformed purely clinical or structural models. CONCLUSION: The cumulative evidence suggests that combinations of structural and clinical risk factors may be able to predict radiographic KOA progression, particularly in patients with accelerated progression. Clinically relevant and feasible prediction models and risk calculators may provide valuable decision-making support when caring for patients at risk of KOA progression, although standardization in modeling and variable identification does not yet exist.

The Influence of Static Load and Sideways Impact Fall on Extramedullary Bone Plates Used to Treat Intertrochanteric Femoral Fracture: A Preclinical Strength Assessment.

Hip fracture accounts for a large number of hospitalizations, thereby causing substantial economic burden. Majority (> 90%) of all hip fractures are associated to sideways fall. Studies on sideways fall usually involve loading at quasi-static or at constant displacement rate, which neglects the physics of actual fall. Understanding femur resonance frequency and associated mode shapes excited by dynamic loads is also critical. Two commercial extramedullary implants, proximal femoral locking plate (PFLP) and variable angle dynamic hip screw (VA-DHS), were chosen to carry out the preclinical assessments on a simulated Evans-I type intertrochanteric fracture. In this study, we hypothesized that the behavior of the implant depends on the loading types-axial static and transverse impact-and a rigid implanted construct will absorb less impact energy for sideways fall. The in silico models were validated using experimental measurements of full-field strain data obtained from a 2D digital image correlation (DIC) study. Under peak axial load of 3 kN, PFLP construct predicted greater axial stiffness (1.07 kN/mm) as opposed to VA-DHS (0.85 kN/mm), although the former predicted slightly higher proximal stress shielding. Further, with greater mode 2 frequency, PFLP predicted improved performance in resisting bending due to sideways fall as compared to the other implant. Overall, the PFLP implanted femur predicted the least propensity to adverse stress intensities, suggesting better structural rigidity and higher capacity in protecting the fractured femur against fall.

A scoping review of transfer learning research on medical image analysis using ImageNet.

OBJECTIVE: Employing transfer learning (TL) with convolutional neural networks (CNNs), well-trained on non-medical ImageNet dataset, has shown promising results for medical image analysis in recent years. We aimed to conduct a scoping review to identify these studies and summarize their characteristics in terms of the problem description, input, methodology, and outcome. MATERIALS AND METHODS: To identify relevant studies, MEDLINE, IEEE, and ACM digital library were searched for studies published between June 1st, 2012 and January 2nd, 2020. Two investigators independently reviewed articles to determine eligibility and to extract data according to a study protocol defined a priori. RESULTS: After screening of 8421 articles, 102 met the inclusion criteria. Of 22 anatomical areas, eye (18%), breast (14%), and brain (12%) were the most commonly studied. Data augmentation was performed in 72% of fine-tuning TL studies versus 15% of the feature-extracting TL studies. Inception models were the most commonly used in breast related studies (50%), while VGGNet was the common in eye (44%), skin (50%) and tooth (57%) studies. AlexNet for brain (42%) and DenseNet for lung studies (38%) were the most frequently used models. Inception models were the most frequently used for studies that analyzed ultrasound (55%), endoscopy (57%), and skeletal system X-rays (57%). VGGNet was the most common for fundus (42%) and optical coherence tomography images (50%). AlexNet was the most frequent model for brain MRIs (36%) and breast X-Rays (50%). 35% of the studies compared their model with other well-trained CNN models and 33% of them provided visualization for interpretation. DISCUSSION: This study identified the most prevalent tracks of implementation in the literature for data preparation, methodology selection and output evaluation for various medical image analysis tasks. Also, we identified several critical research gaps existing in the TL studies on medical image analysis. The findings of this scoping review can be used in future TL studies to guide the selection of appropriate research approaches, as well as identify research gaps and opportunities for innovation.

Natural language processing with deep learning for medical adverse event detection from free-text medical narratives: A case study of detecting total hip replacement dislocation.

BACKGROUND: Accurate and timely detection of medical adverse events (AEs) from free-text medical narratives can be challenging. Natural language processing (NLP) with deep learning has already shown great potential for analyzing free-text data, but its application for medical AE detection has been limited. METHOD: In this study, we developed deep learning based NLP (DL-NLP) models for efficient and accurate hip dislocation AE detection following primary total hip replacement from standard (radiology notes) and non-standard (follow-up telephone notes) free-text medical narratives. We benchmarked these proposed models with traditional machine learning based NLP (ML-NLP) models, and also assessed the accuracy of International Classification of Diseases (ICD) and Current Procedural Terminology (CPT) codes in capturing these hip dislocation AEs in a multi-center orthopaedic registry. RESULTS: All DL-NLP models outperformed all of the ML-NLP models, with a convolutional neural network (CNN) model achieving the best overall performance (Kappa = 0.97 for radiology notes, and Kappa = 1.00 for follow-up telephone notes). On the other hand, the ICD/CPT codes of the patients who sustained a hip dislocation AE were only 75.24% accurate. CONCLUSIONS: We demonstrated that a DL-NLP model can be used in largescale orthopaedic registries for accurate and efficient detection of hip dislocation AEs. The NLP model in this study was developed with data from the most frequently used electronic medical record (EMR) system in the U.S., Epic. This NLP model could potentially be implemented in other Epic-based EMR systems to improve AE detection, and consequently, quality of care and patient outcomes.

Comparing the performance of a deep convolutional neural network with orthopedic surgeons on the identification of total hip prosthesis design from plain radiographs.

PURPOSE: A crucial step in the preoperative planning for a revision total hip replacement (THR) surgery is the accurate identification of the failed implant design, especially if one or more well-fixed/functioning components are to be retained. Manual identification of the implant design from preoperative radiographic images can be time-consuming and inaccurate, which can ultimately lead to increased operating room time, more complex surgery, and increased healthcare costs. METHOD: In this study, we present a novel approach to identifying THR femoral implants' design from plain radiographs using a convolutional neural network (CNN). We evaluated a total of 402 radiographs of nine different THR implant designs including, Accolade II (130 radiographs), Corail (89 radiographs), M/L Taper (31 radiographs), Summit (31 radiographs), Anthology (26 radiographs), Versys (26 radiographs), S-ROM (24 radiographs), Taperloc Standard Offset (24 radiographs), and Taperloc High Offset (21 radiographs). We implemented a transfer learning approach and adopted a DenseNet-201 CNN architecture by replacing the final classifier with nine fully connected neurons. Furthermore, we used saliency maps to explain the CNN decision-making process by visualizing the most important pixels in a given radiograph on the CNN's outcome. We also compared the CNN's performance with three board-certified and fellowship-trained orthopedic surgeons. RESULTS: The CNN achieved the same or higher performance than at least one of the surgeons in identifying eight of nine THR implant designs and underperformed all of the surgeons in identifying one THR implant design (Anthology). Overall, the CNN achieved a lower Cohen's kappa (0.78) than surgeon 1 (1.00), the same Cohen's kappa as surgeon 2 (0.78), and a slightly higher Cohen's kappa than surgeon 3 (0.76) in identifying all the nine THR implant designs. Furthermore, the saliency maps showed that the CNN generally focused on each implant's unique design features to make a decision. Regarding the time spent performing the implant identification, the CNN accomplished this task in ~0.06 s per radiograph. The surgeon's identification time varied based on the method they utilized. When using their personal experience to identify the THR implant design, they spent negligible time. However, the identification time increased to an average of 8.4 min (standard deviation 6.1 min) per radiograph when they used another identification method (online search, consulting with the orthopedic company representative, and using image atlas), which occurred in about 17% of cases in the test subset (40 radiographs). CONCLUSIONS: CNNs such as the one developed in this study can be used to automatically identify the design of a failed THR femoral implant preoperatively in just a fraction of a second, saving time and in some cases improving identification accuracy.

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

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.