Intraobserver and interobserver agreement, reproducibility, and accuracy of computed tomographic measurements of pituitary gland dimensions in healthy dogs

Abstract

Objective - To determine the reproducibility and accuracy of computed tomographic (CT) measurements of pituitary gland dimensions in healthy dogs. Animals - 35 healthy sexually intact adult dogs. Procedures - 2 observers independently viewed CT images of the skull in 35 dogs twice. Pituitary gland height, width, length, and volume and pituitary gland height-to-brain area ratio (P:B ratio) were measured or calculated. Intraobserver and interobserver agreement indexes (Als) were calculated for pituitary gland dimensions. Computed tomography was performed also on 5 phantoms, and both observers measured phantom dimensions twice. True-value Als were calculated for the phantom study. Results - The mean ± SD interobserver Al between observer 1 and 2 for pituitary gland height and the P:B ratio was 0.90 ± 0.07 The intraobserver Al for pituitary gland height and the P:B ratio was 0.97 ± 0.04 for observer 1 and 0.94 ± 0.04 for observer 2. The intra- and interobserver Als for the other dimensions were lower than those for pituitary gland height and the P:B ratio. All phantom dimensions on CT images were underestimated significantly, compared with their true values. Conclusions and Clinical Relevance - The intra- and interobserver Als for pituitary gland dimension measurements on CT images were high. However, the same observer preferably should perform serial measurements. Window settings influence pituitary gland dimension measurements, and predetermined window settings are recommended to make comparisons among dogs. Pituitary gland dimension measurements made from CT images in our study underestimated the true values.