Digital Analysis of Subtrochlear Sclerosis in Elbows Submitted for Dysplasia Screening

Ulnar trochlear notch (UTN) subchondral bone sclerosis is observed in elbow dysplasia (ED) associated with the medial coronoid disease. However, its evaluation is based on a simple visual examiner assessment of bone radio-opacity level and is considered subjective. The purpose of this study was to objectively characterize the radiographic opacity of the ulnar trochlear notch (UTN) subchondral bone in mediolateral elbow projections classified, using the International Elbow Working Group guidelines. Records and mediolateral flexed elbow images from the Danish Kennel Club database for the ED screening scheme between 2012 and 2018 were available. Of the dogs in the database, those with an ED-negative status in the left limb were identified. From these, 20 dogs each having a status free from ED, or with Grade 1, 2, or 3 in the right limb, were randomly chosen. Joints with primary ununited anconeal process were excluded from the sample. A template was developed using the ImageJ software, for computer UTN sclerosis analysis. It was overlaid onto each image to define five regions of interest (ROIs): ROI-1, distal UTN; ROI-2, middle UTN; ROI-3, caudal UTN; ROI-4, cortical bone; and ROI-5, bone marrow. Mean pixel intensity for each UTN ROI was divided by the mean pixel intensity of ROI-4 to normalize the data. The mean ± standard deviation (SD) of the normalized pixel intensity in the disease joints (ED Grades 1, 2, and 3) was 1.18 ± 0.17, 1.03 ± 0.12, and 0.92 ± 0.09 for ROIs 1, 2, and 3, respectively. The corresponding values for the contralateral normal left joints were 1.16 ± 0.17, 1.01 ± 0.1, and 0.91 ± 0.08. There was a significant difference (P < 0.05) in the normalized mean pixel intensity in dysplastic vs. non-dysplastic elbow joints for ROIs 1 and 2. The raw mean pixel intensity from right and left cortical and marrow bone ROIs sometimes showed relatively large differences. Digital radiography is associated with exposure and post-processing variabilities. Differences in apparent radio-opacity (as indicated by pixel intensity) though statistically significant in dysplastic joints compared with contralateral normal joints are slight and are thus problematic for computer-aided assessments of UTN sclerosis.