We aimed to analyse the clinical outcomes and survivorship of anatomic total shoulder arthroplasty using a stemless humeral component with cemented pegged polyethylene glenoid performed with the technique of eccentric reaming to partially correct retroversion. These results were then compared with TSA using the same implant for end-stage shoulder arthritis with a normal version of the native glenoid. A retrospective case series was performed using a prospectively collected database of anatomic TSA patients operated at Woodend General Hospital, Aberdeen, UK. Between 2010 and 2019, 107 total shoulder arthroplasties (TSA) were done using standard anatomic stemless TSA implants (Affinis Short, Mathys Ltd, Bettlach, Switzerland) in 98 patients. Standardized preoperative and postoperative shoulder radiological imaging for glenoid retroversion was collected. Depending on the angle of native glenoid version, patients were divided into retroverted and non-retroverted glenoid as per the Walch Classification. To assess the radiological outcome at the final follow-up, radiolucency was assessed on the glenoid and humeral side using the Lazarus grading. The final clinical and radiologic outcome from the retroverted group was compared with the population with a non-retroverted glenoid. Five TSAs were excluded from the analysis as they did not have satisfactory postoperative radiographs. Hence, a total of 102 shoulders were available for analysis.Objective
Design and methods
Technological advancements in orthopaedic surgery have mainly focused on increasing precision during the operation however, there have been few developments in post-operative physiotherapy. We have developed a computer vision program using machine learning that can virtually measure the range of movement of a joint to track progress after surgery. This data can be used by physiotherapists to change patients’ exercise regimes with more objectively and help patients visualise the progress that they have made. In this study, we tested our program's reliability and validity to find a benchmark for future use on patients. We compared 150 shoulder joint angles, measured using a goniometer, and those calculated by our program called ArmTracking in a group of 10 participants (5 males and 5 females). Reliability was tested using adjusted R squared and validity was tested using 95% limits of agreement. Our clinically acceptable limit of agreement was ± 10° for ArmTracking to be used interchangeably with goniometry. ArmTracking showed excellent overall reliability of 97.1% when all shoulder movements were combined but there were lower scores for some movements like shoulder extension at 75.8%. There was moderate validity shown when all shoulder movements were combined at 9.6° overestimation and 18.3° underestimation. Computer vision programs have a great potential to be used in telerehabilitation to collect useful information as patients carry out prescribed exercises at home. However, they need to be trained well for precise joint detections to reduce the range of errors in readings.
