Patient Positioning Assistive Technology Applicable to the Existing Computed Tomography System: Estimation by Pixel...

This study aimed to propose a patient positioning assistive technique using computed tomography (CT) scout images. A total of 210 patients who underwent CT scans in a single center, including on the upper abdomen, were divided into a study set of 127 patients for regression and 83 patients for verification. Linear regression analysis was performed to determine the R2 coefficient and the linear equation related to the mean pixel value of the scout image and ideal table height (TH ideal ). The average pixel values of the scout image were substituted into the regression equation to estimate the TH ideal . To verify the accuracy of this method, the distance between the estimated table height (TH est ) and TH ideal was measured. The medians of age (in years), gender (male/female), height (in centimeters), and body weight (in kilograms) for the regression and verification groups were 68 versus 70, 85/42 versus 55/28, 163.8 versus 163.0, and 59.9 versus 61.9, respectively. Linear regression analysis indicated a high coefficient of determination ( R2 = 0.91) between the mean pixel value of the scout image and TH ideal . The correlation coefficient between TH ideal and TH est was 0.95 (95% confidence interval, 0.92-0.97; P < 0.0001), systematic bias was 0.2 mm, and the limits of agreement were -5.4 to 5.9 ( P = 0.78). The offset of the table height with TH est was 2.8 ± 2.1 mm. The proposed estimation method using scout images could improve the automatic optimization of table height in CT, and it can be used as a general-purpose automatic positioning technique.