Determination of optimal window width and level for measurement of the canine pituitary gland height on computed tomographic images using a phantom.

The purpose of this study was to determine the optimal window level and width for accurate measurement of pituitary gland height in a phantom. A phantom, mimicking the appearance of the pituitary gland, was made by using the base of a canine skull immersed in buttermilk. Seven different-sized cylindrical pieces of cheese were made and the real diameter of each sample was measured by using a caliper. The cheese was then placed in the pituitary fossa, and 2-mm thick contiguous computed tomography (CT) slices were acquired. The height of the cheese was measured using electronic calipers with different window width and level (WW-WL) settings. The mean difference between CT and manual caliper measurements and concordance correlation coefficients were calculated for all settings. CT measurements underestimated manual caliper measurements with all settings, but the smallest difference (0.5 mm) was obtained with a 250-80 window. Concordance coefficients ranged from 0.699 to 0.820, with the maximum concordance coefficient obtained with a 250-80 window. A second series of measurements using a different window setting for the brain-pituitary gland edge (WW-WL: 250-80) and for the bone-pituitary gland edge (WW-WL: 250-240) resulted in a reduction in the mean difference (0.1 mm) and an increase in the concordance coefficient (0.836). It was concluded that underestimation of the height of the pituitary gland on CT images is most likely due to erroneous placement of the edge of the basisphenoid bone using a low level and a narrow window.