Shoulder surgeons are trying to understand the clinical importance of preoperative shoulder pathoanatomy and its modification by arthroplasty. Large-scale multicenter studies are necessary for investigating the relationship between standardized preoperative and sequential postoperative anatomical measurements and the outcome realized by the patient.
The validity of such studies will depend on (1) standardization of the measurement methods across different centers, (2) human-observer independency avoiding the risk of inter-observer variation and observer bias, and (3) highly efficient methods that enable the evaluation of very large numbers of images. While CT scans are commonly used for characterizing preoperative shoulder anatomy, they are impractical for evaluating the postoperative anatomy and changes over time. Standardized anteroposterior and axillary radiographs provide a practical and cost-effective approach for making preoperative and postoperative measurements using the same imaging method.
A recent study, Can computer vision / artificial intelligence locate key reference points and make clinically relevant measurements on axillary radiographs? demonstrated the potential of artificial intelligence in assessing clinically important relationships on standardized axillary x-rays. Standardized pre and post arthroplasty axillary radiographs were manually annotated by shoulder surgeons locating six reference points as shown the figure below:
The anterior and posterior edges of the glenoid face are indicated by the green dots.
The center of the glenoid face by the blue dot.
The base of the glenoid vault by the yellow dot.
The spinoglenoid notch by black dot at tip of arrow.
The circle fitting the humerus articular surface by the blue circle.
These points were then used to measure glenoid version and humeroglenoid alignment (HGA-AP). Version was measured as the angle between the red and green lines. HGA-AP was measured as the perpendicular distance (double headed arrow) between the centre of the circle (star) and the perpendicular bisector (yellow line) of the glenoid face line (red line) divided by the diameter of the circle (dotted white line)
These annotated images were used to train a computer vision model that could identify these reference points and determine humeroglenoid alignment in the anterior to posterior direction and glenoid version without human guidance.
The model's accuracy was tested on a separate test set of 52 axillary images that were not used in training the model, comparing the model's reference point locations, humero-glenoid alignment and glenoid version to the corresponding values assessed by the two surgeons.
The model performed efficiently, allowing the rapid uploading of images and analysis of reference points, glenoid version, and humeroglenoid alignment (HGA-AP) without human participation. The model was able to produce the measurements in a matter of seconds compared to approximately two hours required for surgeon assessment of the relatively small set of 52 images.
The model was able to rapidly identify all six reference point locations to within a mean of 2 mm of the surgeon-assessed points. The mean variation in alignment and version measurements between the surgeon assessors and the model was similar to the variation between the two surgeon assessors.
The average differences between the surgeon- and the model-assessed reference points for the test set are shown below
The mean differences in glenoid version and HGA-AP between the surgeon assessors, between each surgeon assessor and the model, and between the average of the two surgeon assessors the model is shown below
The inter-observer variability between the two surgeons was similar to that between the average of the two surgeons and model
While it will require substantial further refinement before it is ready for broad scale application, this proof-of-principle study does demonstrate the development and validation of a computer vision/artificial intelligence model that can independently identify key landmarks and determine the glenohumeral relationship and glenoid version on axillary radiographs. This observer-independent approach has the potential to enable efficient assessment of shoulder radiographs, substantially lessening the burden of manual x-ray interpretation and enabling scaling of these measurements across large numbers of patients from multiple centers so that pre- and postoperative anatomy can be correlated with patient reported clinical outcomes.
Other studies have reported the use of artificial intelligence to classify shoulder implants
Classifying shoulder implants in X-ray images using deep learning "In a data set containing X-ray images of shoulder implants from 4 manufacturers and 16 different models, deep learning is able to identify the correct manufacturer with an accuracy of approximately 80% in 10-fold cross validation, while other classifiers achieve an accuracy of 56% or less"
Development of a machine learning algorithm to identify total and reverse shoulder arthroplasty implants from X-ray images. "This proof of concept study demonstrates that machine learning can assist with preoperative planning and improve cost-efficiency in shoulder surgery."
A novel hybrid machine learning based system to classify shoulder implant manufacturers. "The proposed hybrid machine learning algorithms achieve the goal of low cost and high performance compared to other studies in the literature."
Deep learning to automatically classify very large sets of preoperative and postoperative shoulder arthroplasty radiographs. "We developed an efficient, accurate, and reliable AI algorithm to automatically identify key imaging features of laterality, imaging view, and implant type in shoulder radiographs. This algorithm represents the first step to automatically classify and organize shoulder radiographs on a large scale in very little time, which will profoundly enrich shoulder arthroplasty registries."
Artificial intelligence for automated identification of total shoulder arthroplasty implants. "A DL model demonstrated excellent accuracy in identifying 22 unique TSA implants from 8 manufacturers. "
Other studies have explored the use of artificial intelligence to make measurements on x-ray images
The Development of a Yolov8-Based Model for the Measurement of Critical Shoulder Angle (CSA), Lateral Acromion Angle (LAA), and Acromion Index (AI) from Shoulder X-ray Images. "The results indicated that automatic measurement methods align with manual measurements with high accuracy and offer an effective alternative for clinical applications".
The acetabularization index: a novel measure of acromial bone loss prior to reverse shoulder arthroplasty. "AI is a reliably measurable tool on radiographs and 2D CT scans"
While others have used artificial intelligence to assess humeral fractures
Clinical validation of artificial intelligence-based preoperative virtual reduction for Neer 3- or 4-part proximal humerus fractures "The AI-based preoperative virtual reduction model showed good performance in the reduction model in proximal humerus fractures with faster working times."
Artificial intelligence versus radiologist in the accuracy of fracture detection based on computed tomography images: a multi-dimensional, multi-region analysis "The optimized AI model improves the diagnostic efficacy in detecting extremity fractures on radiographs, and the optimized AI model is significantly better than radiologists in detecting avulsion fractures, "
From the foregoing it is evident that we are on the forefront of the application of computer vision/artificial intelligence to enhance clinically important shoulder research.
Stay tuned!
Here are some videos that are of shoulder interest
Shoulder arthritis - what you need to know (see this link).
How to x-ray the shoulder (see this link).
The ream and run procedure (see this link).
The total shoulder arthroplasty (see this link).
The cuff tear arthropathy arthroplasty (see this link).
The reverse total shoulder arthroplasty (see this link).
The smooth and move procedure for irreparable rotator cuff tears (see this link).
Shoulder rehabilitation exercises (see this link).
