Differentiation between inflammatory and neoplastic orbital conditions based on computed tomographic signs.

Abstract

OBJECTIVE: To identify computed tomographic (CT) signs that could be used to differentiate inflammatory from neoplastic orbital conditions in small animals. ANIMAL STUDIED: Fifty-two animals (25 cats, 21 dogs, 4 rabbits, and 2 rodents). PROCEDURE: Case-control study in which CT images of animals with histopathologic diagnosis of inflammatory (n = 11), neoplastic orbital conditions (n = 31), or normal control animals (n = 10) were reviewed independently by five observers without the knowledge of the history or diagnosis. Observers recorded their observations regarding specific anatomical structures within the orbit using an itemized form containing the following characteristics: definitely normal; probably normal; equivocal; probably abnormal; and definitely abnormal. Results were statistically analyzed using Fleiss' kappa and logistic regression analyses. RESULTS: The overall level of agreement between observers about the presence or absence of abnormal CT signs in animals with orbital disease was poor to moderate, but was highest for observations concerning orbital bones (κ = 0.62) and involvement of the posterior segment (κ = 0.52). Significant associations between abnormalities and diagnosis were found for four structures: Abnormalities affecting orbital bones (odds ratio [OR], 1.7) and anterior ocular structures (OR, 1.5) were predictive of neoplasia, while abnormalities affecting extraconal fat (OR, 1.7) and skin (OR, 1.4) were predictive of inflammatory conditions. CONCLUSIONS: Orbital CT is an imaging test with high specificity. Fat stranding, a CT sign not previously emphasized in veterinary medicine, was significantly associated with inflammatory conditions. Low observer agreement probably reflects the limited resolution of CT for small orbital structures.