Variation in CT Number and Image Noise Uniformity According to Patient Positioning in MDCT

Many algorithms for clinical decision making rely on assessment of the CT number (expressed as Hounsfield units); however, to our knowledge, few, if any, studies have addressed how CT numbers change as a function of patient positioning within the scanner.

An anthropomorphic phantom underwent imaging with varying amounts of vertical orientation misalignment with respect to isocenter. CT number and noise were measured using ROIs in the upper thorax, mid thorax, and abdomen. The degree of noise nonuniformity and changes in the CT number were assessed by comparing values obtained in the anterior versus posterior ROIs. To add clinical relevance, data on vertical mispositioning were collected from 20,316 clinical abdominal CT scans. Box-and-whisker plot analysis was used to identify the range of patient positioning.

Absolute CT number changes of more than 20 HU were observed for some ROIs at phantom positions of 10 cm from isocenter, with important differences noted between the thoracic and abdominal regions. Noise uniformity varied by more than twofold for all regions at 10 cm below isocenter. On clinical CT examinations, off-centering of more than 1, 2, 4, and 6 cm occurred for 41%, 19%, 1.9%, and 0.3% of patients, respectively.

Radiologists should treat CT number measurements with caution when patients are grossly mispositioned in the scanner. The substantial changes in attenuation values shown in the present study are large enough to warrant further investigation.