Pascal Boileau1, Damien Cheval2, Marc-Olivier Gauci1, Nicolas Holzer3, Jean Chaoui4, Gilles Walch5. 1. iULS (Institut Universitaire Locomoteur & du Sport), Hôpital Pasteur 2, University of Nice Sophia-Antipolis, Nice, France. 2. Department of Orthopaedic Surgery, Centre Hospitalier Intercommunal de Cornouaille, Quimper, France. 3. Division of Orthopaedics and Trauma Surgery, Department of Surgery, Geneva University Hospitals, Geneva, Switzerland. 4. Telecom Bretagne and Imascap, Brest, France. 5. Centre Orthopédique Santy, Hôpital Privé Jean Mermoz, Lyon, France.
Abstract
BACKGROUND: Preoperative computed tomography (CT) measurements of glenoid version and inclination are recommended for planning glenoid implantation in shoulder arthroplasty. However, current manual or semi-automated 2-dimensional (2D) and 3-dimensional (3D) methods are user-dependent and time-consuming. We assessed whether the use of a 3D automated method is accurate and reliable to measure glenoid version and inclination in osteoarthritic shoulders. METHODS: CT scans of osteoarthritic shoulders of 60 patients scheduled for shoulder arthroplasty were obtained. Automated, surgeon-operated, image analysis software (Glenosys; Imascap) was developed to measure glenoid version and inclination. The anatomic scapular reference planes were defined as the mean of the peripheral points of the scapular body as well as the plane perpendicular to it, passing along the supraspinatus fossa line. Measurements were compared with those obtained using previously described manual or semi-automated methods, including the Friedman version angle on 2D CTs, Friedman method on 3D multiplanar reconstructions (corrected Friedman method), Ganapathi-Iannotti and Lewis-Armstrong methods on 3D volumetric reconstructions (for glenoid version), and Maurer method (for glenoid inclination).The mean differences (and standard deviation) and the concordance correlation coefficients (CCCs) were calculated. Two orthopaedic surgeons independently examined the images for the interobserver analysis, with one of them measuring them twice more for the intraobserver analysis; interobserver and intraobserver reliability was calculated using the intraclass correlation coefficients (ICCs). RESULTS: The mean difference in the Glenosys glenoid version measurement was 2.0° ± 4.5° (CCC = 0.93) compared with the Friedman method, 2.5° ± 3.2° (CCC = 0.95) compared with the corrected Friedman method, 1.5° ± 4.5° (CCC = 0.94) compared with the Ganapathi-Iannotti method, and 1.8° ± 3.8° (CCC = 0.95) compared with the Lewis-Armstrong method. There was a mean difference of 0.2° ± 4.7° (CCC = 0.78) between the inclination measurements made with the Glenosys and Maurer methods. The difference between the overall average 2D and 3D measurements was not significant (p = 0.45). CONCLUSIONS: Use of fully automated software for 3D measurement of glenoid version and inclination in arthritic shoulders is reliable and accurate, showing excellent correlation with previously described manual or semi-automated methods. CLINICAL RELEVANCE: The use of automated surgeon-operated image analysis software to evaluate 3D glenoid anatomy eliminates interobserver and intraobserver discrepancies, improves the accuracy of preoperative planning for shoulder replacement, and offers a potential gain of time for the surgeon.
BACKGROUND: Preoperative computed tomography (CT) measurements of glenoid version and inclination are recommended for planning glenoid implantation in shoulder arthroplasty. However, current manual or semi-automated 2-dimensional (2D) and 3-dimensional (3D) methods are user-dependent and time-consuming. We assessed whether the use of a 3D automated method is accurate and reliable to measure glenoid version and inclination in osteoarthritic shoulders. METHODS: CT scans of osteoarthritic shoulders of 60 patients scheduled for shoulder arthroplasty were obtained. Automated, surgeon-operated, image analysis software (Glenosys; Imascap) was developed to measure glenoid version and inclination. The anatomic scapular reference planes were defined as the mean of the peripheral points of the scapular body as well as the plane perpendicular to it, passing along the supraspinatus fossa line. Measurements were compared with those obtained using previously described manual or semi-automated methods, including the Friedman version angle on 2D CTs, Friedman method on 3D multiplanar reconstructions (corrected Friedman method), Ganapathi-Iannotti and Lewis-Armstrong methods on 3D volumetric reconstructions (for glenoid version), and Maurer method (for glenoid inclination).The mean differences (and standard deviation) and the concordance correlation coefficients (CCCs) were calculated. Two orthopaedic surgeons independently examined the images for the interobserver analysis, with one of them measuring them twice more for the intraobserver analysis; interobserver and intraobserver reliability was calculated using the intraclass correlation coefficients (ICCs). RESULTS: The mean difference in the Glenosys glenoid version measurement was 2.0° ± 4.5° (CCC = 0.93) compared with the Friedman method, 2.5° ± 3.2° (CCC = 0.95) compared with the corrected Friedman method, 1.5° ± 4.5° (CCC = 0.94) compared with the Ganapathi-Iannotti method, and 1.8° ± 3.8° (CCC = 0.95) compared with the Lewis-Armstrong method. There was a mean difference of 0.2° ± 4.7° (CCC = 0.78) between the inclination measurements made with the Glenosys and Maurer methods. The difference between the overall average 2D and 3D measurements was not significant (p = 0.45). CONCLUSIONS: Use of fully automated software for 3D measurement of glenoid version and inclination in arthritic shoulders is reliable and accurate, showing excellent correlation with previously described manual or semi-automated methods. CLINICAL RELEVANCE: The use of automated surgeon-operated image analysis software to evaluate 3D glenoid anatomy eliminates interobserver and intraobserver discrepancies, improves the accuracy of preoperative planning for shoulder replacement, and offers a potential gain of time for the surgeon.
Authors: Lawrence Lo; Scott Koenig; Natalie L Leong; Brian B Shiu; S Ashfaq Hasan; Mohit N Gilotra; Kenneth C Wang Journal: Skeletal Radiol Date: 2020-10-23 Impact factor: 2.199