Hounsfield units are a useful predictor of pleural effusion cytological type in dogs but not in cats.

All categories of pleural effusion subjectively display as soft tissue opacity on computed tomography (CT). Quantitative measurement using Hounsfield units (HU) has the potential to bring additional information regarding the nature of the fluid in a noninvasive way. The purposes of this retrospective cross-sectional analytical study were to compare Hounsfield units of different pleural effusion categories in dogs and cats, assess association between specific cytologic parameters and Hounsfield units, and evaluate the effect of dependent vs. nondependent aspect of the effusion pool on Hounsfield unit. A total of 111 patients (74 dogs and 37 cats) with pleural effusion, that underwent thoracic CT and diagnostic thoracocentesis, were included in the study. Effusions were cytologically categorized as exudate, transudate, modified transudate, hemorrhage, or chyle. Significant differences existed in Hounsfield units between categories in dogs (P < 0.0001) but not in cats (P = 0.334). Canine chylous effusion (6.1 ± 4.7 HU (mean ± standard deviation)) and transudate (5.6 ± 2.0) were significantly lower than exudate (20.3 ± 9.5) and hemorrhage (21.4 ± 9.2). No significant differences were found between modified transudate (13.6 ± 10.3) and other categories. Significant, weak linear correlation was identified in dogs between Hounsfield units and total protein (P = 0.018, R2  = 0.089), red blood cells (P = 0.021, R2  = 0.077), and total nucleated cells (P = 0.013, R2  = 0.089). The Hounsfield units of dependent effusion was not significantly higher than the nondependent effusion, except for canine chylous effusion (P = 0.008). Fourteen Hounsfield units was identified as the most clinically useful threshold: <14 HU identified transudate or chylous effusion with a sensitivity of 100% and a specificity of 69%. A threshold >14 HU had a specificity of 100% and a sensitivity of 69% for identifying exudate, modified transudate, or hemorrhage.