Download Glenoid Version Measurement when the Medial Scapula Is Not in

Glenoid Version Measurement when the Medial Scapula Is Not in the Cross-Sectional Imaging Field of View
+1,2Robertson, D D; 1,3Sharma, G B; 4Kang, M D; 5Stone, D A; 1,5McMahon, P J
University of Pittsburgh, Bioengineering, USA; +2Emory Spine and Orthopaedic Center, USA; 3University of Calgary, Mechanical and Materials
Engineering, Canada; 4University of Pittsburgh Medical Center, Radiology, USA; University of Pittsburgh Medical Center, Orthopaedic Surgery, USA
[email protected]
1
Introduction: Glenoid version, glenoid surface orientation in the axial
plane, has an important role in the passive stabilization of the articular
components of the glenohumeral joint1. Glenoid shape and version has
been associated with glenohumeral instability2, arthritis3, and rotator cuff
tears4. Cross-sectional computed tomography (CT) and magnetic
resonance (MR) imaging have been demonstrated to be more accurate
than radiography for measuring glenoid version5. Glenoid version
calculation requires a transverse scapular axis, often defined as a line
connecting the glenoid fossa midpoint with the scapular spine medial
endpoint6. Other transverse axis definitions include a line perpendicular
to the glenoid surface midpoint or a line tangent to either the posterior or
anterior glenoid rim7. Prior studies require identification of scapula's
medial border to measure glenoid version. Clinical shoulder joint crosssectional imaging frequently uses a reduced field of view (FOV) without
visualizing scapula's medial border. Thus preventing replication of the
published methods. No method is currently available to measure glenoid
version in routine shoulder cross-sectional imaging studies where the
medial scapular border is not in the FOV.
The purpose of this study was to design and validate a method to
accurately measure glenoid version on cross sectional clinical CT
shoulder images, where the FOV does not include the entire scapula.
Materials and Methods: Forty unembalmed, fresh scapulae cadavers
(20 pairs, 10 male, 10 female, mean age: 52±18 years, mean height:
171±12 cm). Exclusion criteria were moderate or severe arthritis, or
bony radiographic abnormalities in any one of the pair. None of the
included donors had any surgical procedure performed on their scapulae
or humeri. This study was approved by the University of Pittsburgh's
Institutional Review Board.
All scapulae were radiographed anteroposteriorly and laterally. Each
scapula was placed in a custom fixture replicating a neutral supine
position and adjusted to minimize/eliminate version measurement
inaccuracies due to scapula rotation. CT was used to validate the new
cross-sectional glenoid version measurement methodology, although the
results can be directly applied to MR imaging as well. The middle third
of the medial edge of the scapula was positioned parallel to the long axis
of the CT table such that the slices were approximately perpendicular to
the glenoid articular surface (Fig. 1). High-resolution volumetric CT
axial images (slice thickness: 1 mm, FOV: 20×20 cm2) were obtained
producing a “patient supine clinical simulation” CT image study.
Fig. 1. Scapula
orientation for crosssectional imaging. Slice
location is selected axial
slice for version
measurement.
3D computer models were generated for each scapula using bone
segmentations (Amira®). The axial slice similar to that passing through
the 3D scapula model’s glenoid center was selected from the “patient
supine clinical simulation” CT images data set.
The selected CT axial slice was used to measure glenoid version on
“full-scapula” and “partial-scapula” FOV images. “Full-scapula” FOV
true glenoid version were measured as the angle between the line joining
the anterior and posterior glenoid margins with the horizontal6 (Fig. 2A).
Fig. 2. (A) “Full-scapula” image true
glenoid version measurement.
(B) “Partial-scapula” image glenoid
version (*) measurement.
To closely resemble the clinical situation, the selected CT images
were cropped to exclude the medial portion. Glenoid version was
measured on the cropped images: line A connected the anterior and
posterior glenoid articular margins, the transverse axis line B connected
the point of intersection of the cortical bone within the scapula neck to
the midpoint of line A, line C was perpendicular to line B. Glenoid
version was the angle between line A and line C and retroverted if the
posterior margin was below line C (Fig. 2B).
The mean and standard deviation for “full” and “partial-scapula”
glenoid version were computed. Differences between gender and
sidedness were tested with two-tailed independent and paired samples ttests (SPSS, Inc., level of significance:0.05).
Original and repeat “full” and “partial-scapula” glenoid version
measurements were made two times by three observers. Each observer
was trained using three additional sample cases not included in the
analysis. Repeated measures ANOVA tested for differences between
“partial-scapula” glenoid version measurements across the three
observers, two occasions, and three observers-two occasions. Test-retest
reliability (precision) was defined as the mean difference between
repeated measurements of “partial-scapula” glenoid version. Inter- and
intra-observer reliability were estimated using intraclass correlation
(ICC) and standard error of measurement (SEM). Accuracy was defined
as the mean difference between “full-scapula” glenoid version (truth)
and “partial-scapula” glenoid version measurements. A correction factor
was calculated and applied to the three observers “partial-scapula”
glenoid version measurements. One-way ANOVA analysis tested for
mean differences between “full” and “partial-scapula” glenoid version
measurements for the original and corrected data.
Results: Overall, males were more retroverted (-3°±3°; p=0.02) than
females (0°±3°). Glenoid version in rights and lefts was approximately
equal (-2°±4°; p = 0.95). “Full-scapula” glenoid version (-0.4°±3°) was
less retroverted than that from the “partial-scapula” image (-7.5°±3°;
p=0.01). “Partial-scapula” glenoid version measurements were reliable.
Table 1 gives ICC and SEM results. Glenoid version measurement
accuracy was 7.1°±4.0°. Adding a 7° correction factor to “partialscapula” glenoid version measurements yielded values not significantly
different than truth (-0.5°±3°; p=0.995).
Table 1. ICC and SEM results.
InterIntra-Observer
Observer
1
2
3
0.7233
0.6349
0.8953
0.9175
ICC
SEM
1.6°
2.2°
0.9°
0.9°
0.0001
0.0111
0.0001
0.0001
P-Value
Discussion: This study described and validated a new method for
accurately measuring glenoid version on “partial-scapula” axial crosssectional images. Scapular positioning to obtain accurate glenoid version
measurement was optimized. As per literature this was needed whether
the FOV includes the full scapula or not3,5,7. Glenoid version varies from
superior to inferior glenoid8. Clinical utilization has chosen the mid
glenoid for measuring. This is the position we selected to test and
validate. As all tested scapulae were “normal”, this technique was not
demonstrated for use in dysplastic or deformed scapulae. CT was more
accessible for our research. Nevertheless, the principles validated here
are applicable to MR imaging. This study demonstrated that true glenoid
version is different than “partial-scapula” glenoid version. However, the
inaccuracy of the partial scapula view version measurements can be
reduced if not completely eliminated by a correction factor.
Significance: To improve spatial resolution and examination of shoulder
structure clinical CT and MR imaging studies frequently exclude the
entire scapula preventing replication of previously published glenoid
version measurements. A novel measurement method is presented to
enable version measurements on “partial-scapula” cross-sectional
imaging studies.
References: [1] Churchill, 2001, J Shoulder Elbow Surg, 10(4). [2]
Wirth, 1994, Clin Orthop Rel Res, 308. [3] Hoenecke, 2010, J Shoulder
Elbow Surg, 19(2). [4] Tetreault, 2004, J Orthop Res, 22(1). [5]
Nyffeler, 2003, J Shoulder Elbow Surg, 12(5). [6] Friedman, 1992, J
Bone Joint Surg Am, 74(7). [7] Rouleau, 2010, J Shoulder Elbow Surg,
19(8). [8] Inui, 2002, Clin Orthop Rel Res, 403.
Poster No. 2222 • ORS 2012 Annual Meeting