Patient-specific predictors of image noise in coronary CT angiography.

BACKGROUND Coronary computed tomography (CT) angiography can be associated with high radiation exposure. Reduction of tube voltage from 120 kV to 100 kV can reduce the dose by up to 40%, but it also increases image noise. OBJECTIVE We aimed to find a patient-specific predictor of image noise to determine the use of reduced tube voltage. METHODS Contrast-enhanced coronary dual-source CT angiography data sets [prospectively electrocardiogram (ECG)-triggered axial and retrospectively ECG-gated spiral acquisition, rotation of 280 milliseconds, 2 × 128 × 0.6 mm collimation, 100 kV, 320 mAs] of 165 patients (age, 54 ± 13 years) for the detection of coronary artery stenoses were analyzed. Image noise was measured in the aortic root. Influence of body weight, height, body mass index, thoracic cross sectional area, as well as the area of the thoracic solid tissue were analyzed. RESULTS Mean image noise in the aorta was 35.1 ± 8.9 HU. Mean dose length product was 207 ± 184 cm · cGy with an average effective dose of 2.9 ± 2.6 mSv. The patient cohort was divided into tertiles according to image noise. Numerous parameters, including BMI and body weight, were significantly different between the highest and lowest tertiles. In multivariable regression analysis, the area of the thoracic solid tissue was the only independent predictor of image noise (P < 0.0001). CONCLUSIONS The area of the thoracic solid tissue at the level of the aortic root predicts image noise and may hence be used for the decision to reduce tube voltage from 120 kV to 100 kV.