The Future of Clinical Active Shoulder Range of Motion Assessment, Best Practice, and Its Challenges: Narrative Review

Abstract

Optimising outcomes after shoulder interventions requires objective shoulder range of motion (ROM) assessments. This narrative review examines video-based pose technologies and markerless motion capture, focusing on their clinical application for shoulder ROM assessment. Camera pose-based methods offer objective ROM measurements, though the accuracy varies due to the differences in gold standards, anatomical definitions, and deep learning techniques. Despite some biases, the studies report a high consistency, emphasising that methods should not be used interchangeably if they do not agree with each other. Smartphone cameras perform well in capturing 2D planar movements but struggle with that of rotational movements and forward flexion, particularly when thoracic compensations are involved. Proper camera positioning, orientation, and distance are key, highlighting the importance of standardised protocols in mobile phone-based ROM evaluations. Although 3D motion capture, per the International Society of Biomechanics recommendations, remains the gold standard, advancements in LiDAR/depth sensing, smartphone cameras, and deep learning show promise for reliable ROM assessments in clinical settings.

Keywords: 2D-pose; accuracy; clinical assessment; narrative review; range of motion; shoulder; shoulder arthroplasty.

Figure 1

Speculative measurement accuracy vs. processing time of several methods to estimate shoulder range of motion. Clinical practice currently utilises diverse pose estimation tools, such as self-assessment, visual-assessment, and the universal goniometer (in orange). Within clinical settings, potentially more precise and consistent video/sensor-based methods (in blue), such as 2D-pose- and 3D-pose-based techniques, hold the potential for an enhanced accuracy of shoulder range of motion estimates. 2D-pose methods are based on a single camera, whereas 3D-pose methods can be either based on a single camera with an infrared, or a near-infrared Light Detecting and Ranging (LiDAR) depth sensor or multiple camera setups. 2D- and 3D-pose methods are coupled with machine/deep learning models that estimate the key body landmarks for estimating shoulder range of motion. In the realm of biomechanics, Inertial Measurement Units (IMUs) provide the orientation information; however, 3D motion capture is considered the gold standard. It surpasses the accuracy of other methods but requires trained staff to operate and process the data. 3D-fluoroscopy is ranked the highest (in green), but this approach involves the exposure to harmful radiation and necessitates the fitting of bony shapes, which consumes time.

Figure 2

Anatomical illustration of the shoulder joint system. The shoulder girdle comprises three true joints and two functional joints, whose coordinated muscle activation enables its wide range of motion. The true anatomical joints are the sternoclavicular, acromioclavicular, and glenohumeral joints. The functional joints include the subacromial space, which facilitates the smooth gliding between the acromion and the rotator cuff, via bursae, and the scapulothoracic articulation, allowing the scapula to glide on the chest wall. This joint organisation underpins the shoulder’s remarkable mobility. Figure from OpenSim (version 4.4) [22].

Figure 3

Active shoulder range of motion assessment. Illustration of shoulder movements commonly evaluated in clinical assessments.

Figure 4

Simple illustration of the two basic approaches of machine/deep learning-based pose estimation. The top row depicts the top-down approach; first, persons are detected and bounded, then within the bounded person, key body landmarks are identified. The bottom row depicts the bottom-up approach; the individual’s body joint positions are estimated first and then subsequently organised to construct distinct poses.

Figure 5

Impact of anatomical frame definition of the thorax (red lines) on estimated thoracohumeral angle. Both red and green dots represent identified body landmarks. For both rows, the thorax is loosely defined by the shoulder and hip markers and the upper arm is defined by the shoulder and elbow key points. The top row depicts the thorax definition based on the average position of the bi-lateral shoulder and hip markers (red dots), and the resulting thorax orientation is highlighted by the vertical red line. The bottom row depicts the thorax definition based on the ipsilateral shoulder and hip markers (red dots); consequently, the thorax orientation estimate (red line) is angled relative to the thorax definition of the top row, and the thoracohumeral angle will be biased and is estimated to be larger in the bottom row than the top row.