The accuracy and repeatability of an automatic 2D–3D fluoroscopic image-model registration technique for determining shoulder joint kinematics
Introduction
The accurate determination of human shoulder joint kinematics is important for understanding shoulder joint pathology and evaluating the surgical treatment of shoulder joint diseases. Numerous prior studies have used various techniques to measure in vivo shoulder kinematics, such as electromagnetic tracking [1], [2], [3], [4], [5], magnetic resonance (MR) imaging [6], [7], [8], [9], [10], radiostereometric analysis (RSA) [11], [12], [13], and optical motion tracking [14], [15], [16], [17], [18]. However, measuring the motion of the scapula and humerus with sub-millimeter levels of accuracy in six-degrees-of-freedom (6DOF) is still a challenging issue in biomedical engineering [19], [20], [21].
Fluoroscopic image-model registration, using a single image or dual images, has increasingly been used to determine in vivo human shoulder joint motions [20], [22], [23], [24], [25], [26], [27]. The critical step in these techniques is the 2D–3D image registration procedure that aligns a 3D-bone model to those of the captured fluoroscopic single image or dual images, of the corresponding target bone. The accuracy of using these techniques has been reported for studying artificial joints [26], [28], [29], and the joints of the knee [30], [31], [32], [33], [34], shoulder [19], [21], [23], ankle [35] and spine [36]. Prior study has shown that the application of the dual plane image-model registration technique is technically challenging [19], [21], [30], [31], single image plane registration is widely in use to determine joint positions in space [22], [23], [28], [34]. However, no study has directly compared the accuracy of using single versus dual fluoroscopic image-model registration techniques to determine in vivo shoulder kinematics.
Recently, an automatic 2D–3D image matching method has been developed and validated for its application to investigate human knee joint kinematics using fluoroscopic images [29], [33]. In this paper, we assess the accuracy and repeatability of using single and dual fluoroscopic images to study the shoulder joint kinematics in 6DOF using an automatic 2D–3D image-model registration technique adapted to the geometry of the shoulder. The accuracy and repeatability of this method in the determination of shoulder joint positions in space was evaluated through a series of in vitro and in vivo tests. The shoulder joint positions reproduced by the automatic 2D–3D image matching method were compared to the positions determined from RSA, taken as the gold standard for position in this study. The repeatability of using this technique to measure in vivo shoulder joint kinematics was demonstrated using four human subjects. Lastly, the optimization performance of the matching algorithm's convergence was assessed in an idealized testing environment, such to avoid systematic error and bias.
Section snippets
2D–3D image matching method
The experimental setup of a dual fluoroscopic image system (DFIS) has been extensively presented in previous studies [19], [26], [27], [31], [32], [33], [35], [36]. In summary, a common imaging zone (approximately 6.5 L in volume) is formed by two fluoroscopes (BV Pulsera®, Philips, Bothell, WA) for stereophotogrammetric analysis (Fig. 1a). The distance between the fluoroscopic intensifier (295 mm in diameter) and the X-ray source is 982 mm. Within a DFIS, a subject can freely move his/her
In vitro testing DFIS to RSA
Using dual fluoroscopic images, a maximum error of 0.16 ± 0.06 mm in translation and 0.58 ± 0.99° in rotation for the humerus was found comparing automatic registration with DFIS to the gold standard RSA (Table 1). Similarly, for the scapula, a maximum error of 0.30 ± 0.04 mm in translation and 0.36 ± 0.13° in rotation was determined. For image-model registration of the humerus using a single fluoroscopic image (F1), the maximum in-plane translational error was 0.26 ± 0.09 mm and 0.60 ± 1.50° in rotation,
Discussion
In this study, we examined the accuracy and repeatability of an automatic 2D–3D image-model registration method using both single and pairs of fluoroscopic images in determining the 6DOF spatial positions of the shoulder joint. The data indicated that for the humerus and scapula, the error in the spatial positions found using a DFIS compared to RSA, was less than 0.30 mm in translation and 0.58° in rotation for all 6DOF. When using a single fluoroscopic image registration technique, the in-plane
Conflict of interest statement
The authors of this manuscript have nothing to disclose that would bias our work.
Acknowledgements
This study was partially supported by the Harvard Shoulder Service at the Massachusetts General Hospital, a grant from National Program on Key Basic Research Project (2011CB707701) and a grant from National Key Technologies R & D Program of China (2011BAF01B03).
References (44)
- et al.
Direct 3-dimensional measurement of scapular kinematics during dynamic movements in vivo
J Shoulder Elbow Surg
(2001) - et al.
Comparison of scapular kinematics between elevation and lowering of the arm in the scapular plane
Clin Biomech (Bristol, Avon)
(2002) - et al.
Three-dimensional scapulothoracic motion during active and passive arm elevation
Clin Biomech (Bristol, Avon)
(2005) - et al.
Scapulohumeral rhythm and associated spinal motion
Clin Biomech (Bristol, Avon)
(2008) - et al.
Three-dimensional scapular kinematics and scapulohumeral rhythm in patients with glenohumeral osteoarthritis or frozen shoulder
J Biomech
(2008) - et al.
Glenohumeral translation during active and passive elevation of the shoulder—a 3D open-MRI study
J Biomech
(2000) - et al.
Image-based RSA. Roentgen stereophotogrammetric analysis based on 2D–3D image registration
J Biomech
(2008) - et al.
Non-invasive determination of coupled motion of the scapula and humerus—an in vitro validation
J Biomech
(2011) - et al.
Measuring dynamic in vivo glenohumeral joint kinematics: technique and preliminary results
J Biomech
(2008) - et al.
The influence of handheld weight on the scapulohumeral rhythm
J Shoulder Elbow Surg
(2008)
Determination of in vivo glenohumeral translation using fluoroscopy and shape-matching techniques
J Shoulder Elbow Surg
In vivo measurement of subacromial space width during shoulder elevation: technique and preliminary results in patients following unilateral rotator cuff repair
Clin Biomech (Bristol, Avon)
In vivo articular cartilage contact at the glenohumeral joint: preliminary report
J Orthop Sci
Validation of a non-invasive fluoroscopic imaging technique for the measurement of dynamic knee joint motion
J Biomech
Determination of in vivo articular cartilage contact areas of human talocrural joint under weightbearing conditions
Osteoarthritis Cartilage
A technique to measure three-dimensional in vivo rotation of fused and adjacent lumbar vertebrae
Spine J
Fast and accurate automated measurements in digitized stereophotogrammetric radiographs
J Biomech
Digital automated RSA compared to manually operated RSA
J Biomech
Comparative accuracy of radiostereometric and optical tracking systems
J Biomech
A new in vivo technique for three-dimensional shoulder kinematics analysis
Skeletal Radiol
Magnetic resonance-based motion analysis of the shoulder during elevation
Clin Orthop Relat Res
Dynamic MR imaging and stress testing in glenohumeral instability: comparison with normal shoulders and clinical/surgical findings
J Magn Reson Imaging
Cited by (48)
Comparison of instantaneous knee kinematics during walking and running
2022, Gait and PostureComparison of glenohumeral joint kinematics between manual wheelchair tasks and implications on the subacromial space: A biplane fluoroscopy study
2022, Journal of Electromyography and KinesiologyAnalysis of forearm rotational motion using biplane fluoroscopic intensity-based 2D–3D matching
2019, Journal of BiomechanicsMRI vs CT-based 2D-3D auto-registration accuracy for quantifying shoulder motion using biplane video-radiography
2019, Journal of BiomechanicsCitation Excerpt :Therefore, there is not a direct relationship between voxel values of the radiography images and 2D projections of MRI. Glenohumeral (GH) kinematic measurement accuracy using CT-based 2D-3D registration has been studied (Bey et al., 2006; Giphart et al., 2012; Massimini et al., 2011; Zhu et al., 2012). However, validation of GH kinematic accuracy using MRI–based registration has never been done, to our knowledge.
- 1
Digital Medical Engineering Laboratory, C249, School of Medicine, Tsinghua University, Beijing 100084, China. Tel.: +86 10 62783631; fax: +86 10 62794377.
- 2
Department of Mechanical Engineering, MIT, 77 Massachusetts Avenue, Cambridge, MA 02139, USA. Tel.: +1 617 643 7279.