The effect of dose reductions on lesion detection in head CT
The purpose of this study was to quantitatively evaluate the effect of reducing radiation dose (i.e., mAs) on lesion detection in head CT examinations. We used a simulation package (Syngo Explorer) to reconstruct 5-mm thick CT images of the brain of one patient pertaining to the centrum semiovale, the basal ganglia, and the sella turcica. Lesion detection was measured using two Alternate Forced Choice (2-AFC) experiments that measure the lesion contrast (I 92%) corresponding to a detection accuracy of 92%. Two observers performed experiments to investigate detection of low contrast lesions with four sizes ranging from 3 mm to 10 mm and at four x-ray beam intensities ranging from 105 mAs to 300 mAs. Results were plotted as log[I 92%] versus log[mAs], and the slopes were measured for each lesion size. Lowering the mAs always reduced lesion detection performance in all images, and for all lesion sizes. Average slopes of the I 92% versus mAs curves were -0.23 for 3 mm lesions,-0.16 for 4.5 mm lesions, and ∼-0.11 for the 7 and 10 mm lesions. For the smallest lesions investigated (3 mm), doubling the x-ray intensity improved lesion detection performance by ∼ 15%, whereas for the largest sized lesions (7 and 10 mm), doubling the tube current improved lesion detection performance by ∼ 7%. The observed improvements in detection performance are markedly lower than predicted by the Rose model where a doubling of the tube current would be expected to improve detection performance by 29% at all lesion sizes. © 2009 SPIE.
