Artificial Intelligence-Based Solution in Personalized Computer-Aided Arthroscopy of Shoulder Prostheses

Abstract

Background: Early recognition of prostheses before reoperation can reduce perioperative morbidity and mortality. Because of the intricacy of the shoulder biomechanics, accurate classification of implant models before surgery is fundamental for planning the correct medical procedure and setting apparatus for personalized medicine. Expert surgeons usually use X-ray images of prostheses to set the patient-specific apparatus. However, this subjective method is time-consuming and prone to errors. Method: As an alternative, artificial intelligence has played a vital role in orthopedic surgery and clinical decision-making for accurate prosthesis placement. In this study, three different deep learning-based frameworks are proposed to identify different types of shoulder implants in X-ray scans. We mainly propose an efficient ensemble network called the Inception Mobile Fully-Connected Convolutional Network (IMFC-Net), which is comprised of our two designed convolutional neural networks and a classifier. To evaluate the performance of the IMFC-Net and state-of-the-art models, experiments were performed with a public data set of 597 de-identified patients (597 shoulder implants). Moreover, to demonstrate the generalizability of IMFC-Net, experiments were performed with two augmentation techniques and without augmentation, in which our model ranked first, with a considerable difference from the comparison models. A gradient-weighted class activation map technique was also used to find distinct implant characteristics needed for IMFC-Net classification decisions. Results: The results confirmed that the proposed IMFC-Net model yielded an average accuracy of 89.09%, a precision rate of 89.54%, a recall rate of 86.57%, and an F1.score of 87.94%, which were higher than those of the comparison models. Conclusion: The proposed model is efficient and can minimize the revision complexities of implants.

Chest X-ray Foreign Objects Detection Using Artificial Intelligence

“…The standard method usually used in the absence of implant documentation, i.e., the comparison of radiographs of the patient with radiographs of the shoulder prosthesis by a specialist, is time consuming and error prone. The results of the IMFC-Net model they developed were promising—an average accuracy of 89.09%, a precision rate of 89.54%, a recall rate of 86.57%, and an F1 score of 87.94% [ 23 ]. However, our model demonstrated the ability to detect shoulder endoprostheses with an accuracy of 98.80%.…”

“…However, AI does not only support diagnostic processes; it can also help when planning and carrying out therapy. Sultan et al proposed three different deep learning-based frameworks to identify different types of shoulder implants in X-ray scans, mainly an ensemble network called the Inception Mobile Fully Connected Convolutional Network (IMFC-Net) [ 23 ]. Only selected examples are listed here, but they illustrate how broadly AI-based algorithms can be used.…”

“…Applications for mobile devices are also being developed to further facilitate the process for clinicians [ 21 ]. Examples of devices a radiologist may encounter include shoulder joint endoprostheses [ 23 , 24 ], bowel motility markers, gastric banding, abdominal tubes [ 25 ], neurostimulator endoprostheses, hemostatic cellulose gauze, vascular bypass [ 26 ], intra-aortic balloon pumps [ 26 ], ventriculoperitoneal valves [ 26 , 27 ], tracheobronchial stents [ 26 ], coronary artery stents [ 26 , 27 ], ventricular assist devices [ 26 , 27 , 28 ], endoscopic capsules, endobronchial devices, devices used in prostate brachytherapy, or hernia meshes [ 25 , 26 ]. In the following study, AI was used to determine selected artifacts on X-ray images.…”

Modern Applications of Machine Learning in Shoulder Arthroplasty

“…Some studies state the reason for this being the wide range over which an anteroposterior shoulder X-ray can be spread 53 . However, many recent studies have reported on ML algorithms that demonstrated accuracy over 90% 53,54,58 . Geng et al conducted a promising proof of concept showing a 93% accuracy and a 0.1 second average identification time 59 .…”