Anish Ghodadra MD, Presenter: Nothing to Disclose

Alex Rothy, Abstract Co-Author: Nothing to Disclose

Aaron Bois, Abstract Co-Author: Nothing to Disclose

Morgan Jones MD, Abstract Co-Author: Nothing to Disclose

Present methods used to estimate glenoid bone loss in recurrent shoulder instability rely on imaging of the contralateral shoulder or simple estimations of glenoid shape (e.g. a circle). The purpose of this work was to develop a robust method to reconstruct the original shape of the anterior glenoid rim following a glenoid defect using statistical shape modeling.

Fifty-eight pairs of human glenoids (age 15–35) from the Hamman-Todd Collection (Cleveland, OH) were digitized using a 3-dimensional laser scanner. Using custom Matlab software, 2-dimensional glenoid contours were generated. The contours were then resampled to have a fixed number of points for all glenoids. Seventy percent (n = 81) of the glenoid contours were then randomly selected as a training set for generation of a statistical shape model using principal component analysis of the covariance matrix of the coordinates along the contours. The remaining 35 glenoid contours were used for model validation. Anterior glenoid defects were simulated in 5% increments in the anterior-posterior direction. The statistical shape model was then fit to the remaining points in the contours using an iterative algorithm seeking to minimize mean error in the contour fitting.

Principal component analysis yielded five major modes of variation in glenoid shape. Mode 1 corresponded to radius of the posterior half of the glenoid. Mode 2 described the radius of the anterior glenoid. Mode 3 corresponded to the size/curvature of the superior glenoid. Mode 4 described the depth of the glenoid notch and Mode 5 described the curvature of the anterior/superior portion of the inferior glenoid rim. The root mean square median error in defect contour reconstruction was 0.95 mm (Quartiles: 0.6 and 1.7) with a 90th percentile of 2.7 mm and a maximum of 3.5 mm. Figure 1 shows four randomly selected glenoids with defects and their reconstructions.

We were able to successfully reconstruct the contours of glenoid defects using a statistical shape model with a relatively small margin of error. This technique could be used to estimate the original contours of glenoid defects thereby aiding in their surgical reconstruction.

This technique could allow accurate estimation of the amount of bone loss which can help predict the failure rate of soft tissue reconstruction and inform the decision between soft tissue and bony reconstruction procedures.  

Ghodadra, A, Rothy, A, Bois, A, Jones, M, Reconstruction of Glenoid Defects Using a Statistical Shape Model.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14018246.html