SEARCH

SEARCH BY CITATION

Keywords:

  • glenoid;
  • total shoulder arthroplasty;
  • shape modeling;
  • glenoid prosthesis;
  • surgical instrumentation

Summary

  1. Top of page
  2. Summary
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

We applied shape modeling and principal component analysis (PCA) to discover glenoid bone structural relationships relevant to improving glenoid prosthesis features, fixation, and instrumentation. Knowledge of external bone morphology guides prosthesis shape and positioning, while internal bone morphology and bone density influence fixation. CT-based modeling defined nonarthritic glenoid subchondral bone surface and internal structure. First and second principal shape components were related to size and density. Reproducible structural parameters and glenoid feature relationships were discovered. Subchondral bone surface was approximated by a circle inferiorly and a triangle superiorly with the circle's center at a reproducible point along a superior-inferior line. Glenoid vault's maximum depth was at the circle's center, and the highest bone density was in posterior glenoid. Glenoid subchondral bone surface version varied from superior to inferior, but not by sex or side. Male subchondral bone surfaces were larger and more retroverted. Even if subchondral bone surfaces are deformed by arthritis, glenoid morphology can be identified by extra-articular landmarks, permitting location of the glenoid center and scapular orientation (glenoid version). Knowledge obtained from this study directs design of novel prosthesis features and instrumentation for use without pre-op CT or computer aided surgery. © 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 32:1471–1478, 2014.

Total shoulder arthroplasty (TSA) requires revision surgery in 2 to 10% of patients within 5 years.[1, 2] A common cause for revision is glenoid prosthesis loosening.[3, 4] Success depends on glenoid prosthesis design, accurate positioning, fixation, soft-tissue balance, and shoulder biomechanics post-surgery.[1-4] Shape modeling with principal component analysis (PCA) of the glenoid (morphology and bone density) should improve structural knowledge and TSA outcomes by discovering novel approaches to prosthesis design, positioning, and instrumentation.

Glenoid morphology has been evaluated with specimens, radiographs, computed tomography (CT), magnetic resonance imaging (MRI), and computer models.[5-8] The glenoid was reported as oval, pear, or inverted comma shaped, broader inferiorly than superiorly.[9-11] Glenoid version has been measured with radiographs, CT, MRI, and computer modeling.[7, 8, 12-15] Optimal prosthesis version is critical as increased retroversion can cause instability and early loosening.[3, 4, 7] Currently, optimum prosthesis version depends on surgical judgment and experience.[16, 17] Glenoid cancellous bone density and architecture vary with location.[18-20] Knowledge of bone density distribution in relation to identifiable landmarks is essential not only for sound prosthesis fixation, but also for location-specific material property assignment for computational simulations.[21] Mathematically defining the glenoid articular surface and extra-articular scapular landmarks could aid in accurate prosthesis positioning, improving outcomes.

We applied shape modeling and PCA to discover glenoid structure relationships relevant for increasing prosthesis longevity through improved design features, fixation, positioning, and instrumentation. CT-based 3D structural analysis defined the glenoid subchondral bone surface, internal glenoid structure, and relevant extra-articular scapular anatomy.

MATERIALS AND METHODS

  1. Top of page
  2. Summary
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Specimen, Imaging, and 3-D Computer Modeling

This study was approved by our Institutional Review Board. Twenty cadaver scapulae pairs, eleven males (50.2 ± 11.8 yrs) and nine females (60 ± 20.5 yrs), were obtained from Midwestern United States donors. Donor height ranged from 149 to 94 cm (170.6 ± 11 cm). Scapulae pairs were excluded if the donor had abnormalities, including moderate or severe arthritis, in either scapula. No donor had shoulder surgical procedures. AP and lateral radiographs and high-resolution volumetric CT axial images (slice thickness: 1 mm, field of view: 20 × 20 cm2, 512 × 512 matrix) were obtained for each pair after placement in a custom fixture ensuring the plane containing the central scapula body and middle 1/3 of the scapula medial border was perpendicular to the image plane. CT images were imported into a visualization and modeling software package (Amira®, FEI Visualization Sciences Group, Burlington, MA) for bone segmentation from which 3D tessellated computer models were generated (Geomagic Design X®, 3D Systems, Morrisville, NC) (Supplemental Fig. 1). All bone segmentations and measurements described below were performed by one observer.

Subchondral Bone Surface Morphology Measurements

Prior to glenoid feature measurement, the scapula vertical reference axis was defined as a line approximating the middle 1/3 of the scapula medial border and contained in a vertical reference plane approximating the body of the scapula. For lateral view measurements, the scapula computer model was aligned such that the viewing direction was both parallel to the vertical reference plane and perpendicular to the vertical reference axis. Lateral view measurements (Fig. 1) included superior-inferior length between supra- and infra-glenoid tubercles (C) and AP width in the inferior glenoid (D) measured perpendicular to C at a point located 1/3 of C's length superior from the infra-glenoid tubercle. This point was named the inferior glenoid circle center. Radial lengths L1h, L2h, L3h, and L4h from the circle center to the inferior glenoid boundary were measured to check for semicircularity of the inferior glenoid. L1h and L3h, and L2h and L4h were 60° and 30° from D, respectively. Superior glenoid AP width (d) measured perpendicular to C (between the anterior boundary notch and the posterior boundary) was recorded. Shortest distances between the glenoid boundary and the acromion (G), and coracoid (F) processes were measured in planes that bisected the portions of the acromion and coracoid processes connecting with the scapula body, respectively. This produced two glenoid boundary points that could serve as additional references along with the inferior glenoid circle center for prosthesis alignment during TSA. Segments E and H, respectively, connected these reference points closest from the acromion and coracoid processes to the circle center. k° and n°, respectively, defined the angles between E and D, and H and C, respectively. Reference variables E, H, k°, and n° locate the circle center.[6] L1 and L2, and L3 and L4, angles of 30° and 60° from C in clockwise and counter-clockwise directions, respectively, defined the inferior glenoid boundary. Base angles t1° (posterior) and t2° (anterior), and height Δht defined a triangle in the superior glenoid having segment d as base and supra-glenoid tubercle as apex. The amount of anterior or posterior rotation of the glenoid was defined as the superior glenoid tilt b° and was measured between C and the vertical reference axis. For the anterior view measurements, the model was oriented such that the vertical reference plane was normal to the viewing direction. Anterior view measurements (Fig. 1) included inferior- (A: between the inferior glenoid boundary at the infra-glenoid tubercle and the anterior notch) and superior- glenoid (B: between the anterior notch and the superior glenoid boundary at the supra-glenoid tubercle) lengths, θ° between A and B, and θ1° between A and the vertical reference axis compared with previously reported glenoid inclination,[22] and θ2° made by scapula inferior-medial border with vertical reference axis. For posterior view measurements, the scapula was aligned such that the vertical reference plane was normal to the viewing direction, and the line passing through the glenoid center and the midpoint of the root of the scapular spine was horizontal (Fig. 1).[8, 23] θ3° was measured between the middle 1/3 of the scapula medial border and the vertical reference axis.

image

Figure 1. Glenoid and scapula external morphology measurements.

Download figure to PowerPoint

Internal Morphology Measurements

Three axial slices from each scapula were analyzed (Fig. 2). The first (Slice 1) was in the superior glenoid at 2/3rd Δht along C inferior from the supra-glenoid tubercle; Slice 2 was half way between the 1st and 3rd slices in the mid-glenoid, and Slice 3 was at the circle center. For each scapula, a horizontal reference axis was defined as the line normal to the vertical reference plane. Approximate geometries were fitted within all axial slices (Fig. 2). Both surface bone glenoid version (m°) (conventional) and subchondral bone glenoid version (a°) were measured with respect to the horizontal reference axis. Based on our CT imaging protocol, measured glenoid version was comparable to Friedman et al. [12]. D′, U, V, W, X, and Y (only Slice 1) were kept within cortical bone. Anterior (r°) and posterior (s°) margin angles were measured in all axial slices, whereas p° and q° were measured in Slice 1.

image

Figure 2. Internal bone morphology parameters through the superior (Slice 1), center (Slice 2), and inferior (Slice 3) glenoid region.

Download figure to PowerPoint

Glenoid Vault Depth and Regional Relative Bone Density

Trigonometric relations were derived from approximate geometries in axial slices to calculate glenoid vault depth perpendicular to D′ at five locations (0.17D′, 0.34D′, 0.50D′, 0.68D′, and 0.85D′) along its length posterior to anterior for Slices 2 and 3, but only at 0.50D′ for Slice 1 (Fig. 3). The locations for measuring the depth were chosen where fixation could be located in future glenoid designs. These locations were equidistant but fractionally farther from the posterior margin as the glenoid has lower depth posteriorly. For each scapula, Slices 2 and 3 were divided into 5, while Slice 1 was divided into 4 regions of interest (Fig. 3). To simplify analysis, we classified regional bone density as Low, Medium, and High, clinically equivalent to <350, 350–650, and >650 HU, respectively.

image

Figure 3. Glenoid vault depths (top row) and bone density (bottom row) in the three selected axial slices. (ROI: region of interest, H: high bone density, M: medium bone density, L: low bone density).

Download figure to PowerPoint

Statistical Analysis

A Principal Component Analysis (PCA) was performed on all the external and internal variables, including donor age and the humeral head radius [24] as an additional measure to account for overall shoulder size (Matlab®, MathWorks Inc., Natick, MA). To confirm and expand PCA findings, two-tailed paired and independent samples t-tests were performed between right and left, and male and female scapulae, respectively, for subchondral bone surface and internal morphology parameters. Linear regression analysis was performed for C, D, D′, d, L1, L2, L3, L4, E, H, A, B, and Δht. One-way ANOVA using L1h, L2h, L3h, L4h, 1/3 C, and ½ D, determined whether inferior glenoid boundary can be approximated by a circle. Mean and standard deviation of glenoid vault depth at multiple locations were computed for the three axial slices. Mean bone density was computed for the regions of interest for the three slices. All additional analyses were performed using the software package SPSS® (SPSS Inc., Chicago, IL) with p < 0.05. To test for accuracy and reliability, lengths C and D and angles b° and θ° were re-measured for 20 randomly selected scapulae. Measurements were made on cadaveric scapulae using precision calipers and a goniometer, and on computer models of scapulae. Accuracy was defined as the mean difference between cadaver and original computer model measurement. Reliability was defined as the mean difference between repeated and original model measurements. Repeatability was a measure of reliability relative to variation among specimens.

RESULTS

  1. Top of page
  2. Summary
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Subchondral Bone Surface Morphology Measurements (Supplemental Table 1)

The superior-inferior glenoid length measurement, C, approximated a normal frequency distribution. (Supplemental Fig. 2) Superior-inferior glenoid length and AP glenoid width, D, were significantly greater in males than females and equal between rights and lefts. AP width in the superior glenoid was equal between males and females and rights and lefts. L1, L2, L3, and L4 were significantly greater in males and equal between rights and lefts. Also, the radius-like measurements, L1h, L2h, L3h, and L4h, were greater in males. The inferior glenoid boundary was a 120° arc subtended at circle center from L2h–L4h, whose center was 1/3 of its length superior from the infra-glenoid tubercle along C. Table 1 lists correlation coefficients between selected measurements. Triangle base angles, t1° and t2°, in superior glenoid were equal in males and females. The posterior base angle, t1°, was significantly greater in lefts than rights. Triangle's height, ▵ht, was greater in males and equal between rights and lefts. Angles k° and n° were equal between sexes and rights and lefts. Thus, the glenoid boundary points closest to the acromion, and the coracoid processes are at approximately constant angles from D and C, respectively. Glenoid tilt, b°, inclination between the inferior and superior glenoid (θ°), and inferior glenoid inclination (θ1°) were equal between sexes and greater in rights than lefts.

Table 1. Corrleation Coefficients and p-values (in parantheses) for Select Glenoid Morphology
Morphology ParametersH (mm)D (mm)D′ (mm)E (mm)Δht (mm)d (mm)A (mm)B (mm)
C (mm)0.92 (0.0)0.80 (0.0)0.72 (0.0)0.51 (0.0)0.73 (0.0)0.63 (0.0)0.62 (0.0)0.66 (0.0)
H (mm) 0.76 (0.0)0.64 (0.0)0.43 (0.0)0.59 (0.0)0.53 (0.0)0.63 (0.0)0.55 (0.0)
D (mm)  0.74 (0.0)0.45 (0.0)0.52 (0.0)0.54 (0.0)0.69 (0.0)0.30 (0.06)
D′ (mm)   0.74 (0.0)0.40 (0.0)0.45 (0.0)0.50 (0.0)0.40 (0.0)
E (mm)    0.23 (0.2)0.41 (0.0)0.23 (0.2)0.43 (0.0)
Δht (mm)     0.73 (0.0)0.47 (0.0)0.53 (0.0)
d (mm)      0.39 (0.0)0.42 (0.0)
A (mm)       0.00 (0.0)

Internal Morphology Measurement: (Supplemental Table 2)

Glenoid version followed an approximately normal frequency distribution (Supplemental Fig. 2). AP width, D′, decreased from the inferior to superior glenoid. D′ was greater in males in the inferior- and mid-glenoid, respectively, and equal in the superior glenoid. D′ was equal between rights and lefts in the inferior glenoid and, in both the mid- and superior-glenoid, greater in rights than lefts. Glenoid anterior wall length, U, decreased from the inferior to superior glenoid. This length was greater in males in the inferior glenoid, and equal in the mid- and superior-glenoid. Glenoid proximal posterior wall length, V, was lowest in the mid-glenoid and greater in males in the superior glenoid. Distal glenoid posterior wall length, W, was greater in males in the inferior glenoid. Glenoid neck width, X, was maximum in the mid-glenoid and greater in males in the inferior-, mid- and superior-glenoid Anterior margin angle, r°, increased from the inferior to superior glenoid, and was equal in the inferior glenoid between sexes and rights and lefts. Posterior margin angle, s° was minimum in mid-glenoid and equal between sexes and rights and lefts in the inferior-, mid-, and superior-glenoid. Both, surface bone glenoid, m°, and subchondral bone glenoid version, a°, were retroverted on average and more retroverted in the superior than inferior glenoid. Males were significantly more retroverted in the inferior- and mid-glenoid than females.

Glenoid Vault Depth and Regional Relative Bone Density

Glenoid vault depth was greater in males at most locations. The maximum depth was at the circle center in the inferior glenoid and was greater in males. Depth was minimum in the posterior mid-glenoid and was greater in males on the anterior-glenoid. Glenoid vault depth was equal between right and left scapulae in all locations, except in the superior glenoid where it was greater in lefts than rights.

Relatively high bone density was present in the posterior-center glenoid, whereas low bone density was noted in the inferior center and superior glenoid regions. Medium bone density was observed in the inferior-anterior glenoid (Fig. 3).

Principal Component Analysis

Specimen score and variable loading plots for the first two principal components are shown in Fig. 4. Supplemental Table 3 shows the eigenvalues for the first four principal components along with the eigenvectors and the variation explained by each component. Normalized data were re-projected along the first four principal components. The 1st principal component was significant in distinguishing between sexes (p < 0.0001), while the 3rd (p = 0.008) and 4th (p = 0.018) were significant in distinguishing right and left specimens. The variables with a majority contribution to the first four principal components are shown in Table 2.

image

Figure 4. Specimen score plot (left) and variable loading plot (right) for the first two principal components.

Download figure to PowerPoint

Table 2. Key Variables Contributing to the First Four Principal Components
Key Vriables Contributing to the Principal Components
IIIIIIIV
H RAD (mm)S1ROI 1 (HU)S1ROI 1 (HU)θ1 (deg)
C (mm)S1ROI 2 (HU)S2U (mm)S1a (deg)
D (mm)S1ROI 4 (HU)S2s (deg)S2r (deg)
L4 (mm)S2ROI 1 (HU)S2H17 (mm)S2H68 (mm)
L3H (mm)S2ROI 2 (HU)S2H34 (mm)S2H85 (mm)
S3D (mm)S2ROI 3 (HU)S2H50 (mm)S3r (deg)
S3U (mm)S2ROI 4 (HU)S3s (deg)S3H68 (mm)
 S2ROI 5 (HU)S3H17 (mm)S3H85 (mm)
 S3ROI 1 (HU)S3ROI 4 (HU) 
 S3ROI 2 (HU)S3ROI 5 (HU) 
 S3ROI 3 (HU)  

Accuracy, Reliability, and Repeatability

Table 3 lists, for parameters C, D, b°, and θ°, the mean, standard deviation, and intraclass correlation coefficient (ICC) for the mean absolute differences between actual scapulae and corresponding computer models (accuracy), and repeated and original scapulae computer models measurements (reliability). Linear measurements were more accurate and reliable compared to angular measurements. ICC was excellent for linear measures (ICC ≥ 0.98) and acceptable for angular measures (ICC ≥ 0.66).

Table 3. Accuracy and Reliability (mean, SD, ICC). Actual Specimen, Computer Model
ParameterMean Absolute Difference (Physical – Model) ± S.DMean Absolute Difference (model) (Repeated – Original) ± S.D.ICC
C (mm)0.7 ± 0.60.5 ± 0.40.98
D (mm)0.6 ± 0.60.7 ± 0.50.98
b (degree)2.6 ± 1.33.6 ± 2.80.66
θ (degree)2.7 ± 0.83.3 ± 2.30.87

DISCUSSION

  1. Top of page
  2. Summary
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

TSAs often require revision due to glenoid prosthesis loosening.[1-4] Glenoid shape and structural knowledge is crucial when improving TSA success through advancement of prosthesis design, positioning, and fixation, soft-tissue balance, and post-surgical shoulder biomechanics. We discovered arthroplasty-relevant glenoid subchondral bone surface and internal structure, as well as scapular anatomy. Our results agreed with prior investigations of glenoid shape. We also identified new external bone morphology that can guide future glenoid prosthesis positioning, internal bone structure that can influence fixation, and both external and internal glenoid bone parameters that can guide future prosthesis designs.

The inferior glenoid boundary is a 120° arc, approximated by a circle, whose center we termed the inferior glenoid circle center. Previously, Burkhart et al.[25] and Huysmans et al.[6] defined the glenoid bare spot as the center of an approximate circle in the inferior glenoid region. However, Kralinger et al.[26] found the glenoid bare spot difficult to locate and occasionally eccentric. Our method finds the center independent of the bare spot.

Arthritis deforms the subchondral bone surface and can make it difficult to find morphology, such as the inferior glenoid center.[15, 27] We found two solutions: one defines the inferior glenoid circle center using an extra-articular reference, the superior-inferior glenoid length, while the second uses the glenoid boundary points closest to the acromion and coracoid process. These two extra-articular methods could accurately locate the circle center even if the subchondral bone surface is deformed, as in the case of arthritis. Work utilizing arthritic glenoids is underway to confirm this.

The superior glenoid was approximated by a triangle. Therefore, the glenoid subchondral bone surface can be approximated using geometric analogs (circle inferiorly, triangle superiorly). Understanding the subchondral bone surface as geometric analogs could improve prosthesis design and may enable surgeons to separately prepare the superior and inferior glenoid with smaller and easier-to-use tools.

Improvement in glenoid prosthesis design and fixation requires improved understanding of glenoid internal structure. Previous work was limited in defining the glenoid vault.[5, 8] We defined glenoid internal structure by quantifying its shape, bone density distribution, and bone depth, parameters, all relevant to peg or keel, cemented or noncemented prostheses.

Gender-specific glenoid designs may have advantages. For example, designs may benefit by gender-specific AP narrowing in the superior glenoid and overall smaller sizes for females. Females also had significantly smaller glenoid neck widths in the superior-, mid-, and inferior- glenoid. This suggests smaller peg diameters and keels for females. For both genders, asymmetric sized pegs may be relevant as glenoid neck width was greatest in the mid-glenoid.

The inferior glenoid anterior margin angle was equal between sexes, and right and lefts. De Wilde et al.[12] found a similar angle between the glenoid anterior side and glenoid subchondral bone surface (coined the Resch angle). Reference to this angle may be relevant to surface preparation and prosthesis positioning.

Glenoid version is important for stability.[2-4, 7] Previous investigations [7, 8, 12, 14, 23, 26] used Friedman et al.'s [13] technique to measure mid-glenoid bone surface version. Our mid-glenoid version measurements were comparable to those in previous investigations.[7, 8, 12, 23] We also measured version in the superior and inferior glenoid and found differences in agreement with prior work.[7, 14]

In TSA the surgeon sees the bone surface as much of the cartilage is gone. We measured glenoid tilt as the rotation of the glenoid surface in the lateral view, glenoid inclination as the superior or inferior angulation of the glenoid surface in the anterior view, and glenoid version as the anterior or posterior angulation of the glenoid surface in several axial cross-sections. Previous investigations measured bone surface version, not the version of the deeper subchondral bone. Our data indicate less superior-inferior difference exists in subchondral bone version than surface version. Future work may explore the relevance of the tilt, inclination, and location varying version as well as subchondral bone version to TSA design and surface preparation.

Glenoid vault depth at the superior-, mid-, and inferior-glenoid were comparable to those found by Anglin et al. [18] We found vault depth maximum at the inferior glenoid circle center and decreased anteriorly, posteriorly, and superiorly. At most locations depth was greater in males, especially the circle center and anterior mid-glenoid, suggesting gender-specific designs. We found high bone density in the posterior mid-glenoid and medium density in the anterior inferior and mid-glenoid. These findings compare favorably with previous work.[18-20] Combining bone density with glenoid vault depth data suggests prioritizing fixation at the circle center and mid-glenoid, with a peg or keel being long enough to engage the high density bone. Our statistical glenoid vault modeling database may be used to test the fit and proximity to dense bone of different prosthesis peg or keel locations, orientations, and lengths. The database can also be used to test design features and their influence on bone stresses and bone remodeling.[21]

Structural modeling and principal component analysis expanded general and arthroplasty-relevant glenoid bone knowledge. The 1st and 2nd principal components were related to size and density and differentiated between sexes. 3rd and 4th principal components' main contributions were from internal structure. The 3rd principal component significantly differentiated between right and left glenoids, possibly due to the dominant and non-dominant arm influences on glenoid internal structure. Glenoid subchondral bone description as a circle inferiorly and triangle superiorly, together with the subchondral bone version data, and the mathematical relationships found between the extra-articular landmarks and the glenoid subchondral bone should aid in surgical preparation of the glenoid (Fig. 5). Our findings may be used to create solutions for positioning the prosthesis in version (relative to the scapula) without the need for pre-operative CT or computer aided surgery (Fig. 5). Glenoid internal bone morphology (vault depth and bone density) can improve prosthesis fixation by selecting peg or keel locations and prosthesis materials that are mechanically and biologically advantageous (Fig. 5). Along with others,[28, 29] ongoing work explores confirmation of these findings in severely deformed glenoids. In summary, we hope these results will improve glenoid prosthesis longevity and stimulate improvement in prosthesis design and fixation, as well as aid in surface preparation and prosthesis positioning during TSA.

image

Figure 5. Glenoid analysis derived prosthesis features and instrumentation. Peg locations, lengths, and orientations (top left), vault center guides independent of glenoid surface (top middle and right), version guides independent of scapula orientation and glenoid surface (bottom left and middle), and glenoid reamers (bottom right). Guides do not require pre-op CT scanning or CAS.

Download figure to PowerPoint

REFERENCES

  1. Top of page
  2. Summary
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Additional supporting information may be found in the online version of this article at the publisher's web-site.

FilenameFormatSizeDescription
jor22696-sm-0001-SupTab_S1.docx15KTable S1: Glenoid external morphology measurements (mean, standard deviation, range). Shaded cells show significant differences, p < 0.05.
jor22696-sm-0001-SupTab_S2.docx19KTable S2: Glenoid internal morphology measurements in the three selected axial slices (mean, standard deviation, and p-value)#. Shaded cells show significant defferences, p < 0.05.
jor22696-sm-0001-SupTab_S3.docx18KTable S3: Eigenvalues, eigenvectors and explained data variation first four principal components.
jor22696-sm-0001-SupFig_S1.tif2502KFigure S1: Glenoid (scapula) model generation.
jor22696-sm-0001-SupFig_S2.tif181KFigure S2: Frequency distribution of glenoid superior-inferior length and glenoid version.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.