Our Institutional Review Board approved the retrospective study; and all CT examinations were clinically indicated. In our institution, prospectively ECG-triggered sequential scanning is used for unenhanced lung CT in free-breathing sedated children to eliminate cardiac motion artifacts and reduce respiratory motion artifacts on axial images. However, respiratory misregistration artifacts persist on coronal and sagittal reformatted images between adjacent sequential imaging slabs. In addition to axial images, coronal and sagittal reformatted images are included in routine image interpretation; hence, there is a need to reduce the motion artifacts on reformatted images. In our institution, lung CT protocol was gradually changed from prospectively ECG-triggered sequential scanning to combined prospectively ECG- and respiratory-triggered sequential scanning to reduce the respiratory misregistration artifacts on reformatted images. Combined prospectively ECG- and respiratory-triggered scan mode was installed in our institution in December 2008. Depending on the patients' condition and examination schedule, prospectively ECG-triggered sequential or spiral scan was used for unenhanced chest CT until March 2011. After March 2011, combined prospectively ECG- and respiratory-triggered scan was used as the primary scan mode for unenhanced chest CT in free-breathing children. In this study, we included pediatric patients who consecutively underwent free-breathing unenhanced lung CT using these two different CT scan protocols. Patients older than 15 years of age were excluded.

In 15 children (median age 19 months, range 6 months–8 years; 7 boys), prospectively ECG-triggered sequential lung CT using a first-generation dual-source scanner (SOMATOM Definition, Siemens Healthcare, Forchheim, Germany) was obtained during free breathing with and without additional respiratory triggering on two different days (median interval 3 months, range 12 days–1 year) between December 2008 and March 2011. Lung CT with ECG triggering was performed first in 10 patients and later in five patients. Disease categories of the patient population comprised congenital pulmonary airway malformation (n = 7), pneumonia (n = 2), Wilms tumor (n = 2), interstitial pneumonitis (n = 1), Langerhans cell histiocytosis (n = 1), repaired total anomalous pulmonary venous connection (n = 1), and pectus excavatum (n = 1). All children were sedated with oral chloral hydrate and intravenous administration of midazolam and/or ketamine was added if necessary. All children were younger than 5 years of age except for an 8-year-old boy who needed sedation for lung CT due to altered mentality caused by hemorrhagic brain lesions. In all children, general anesthesia, intubation, or controlled ventilation was not used.

Dual-source prospective ECG-triggered sequential CT scan was performed at a gantry rotation time of 0.33 second (temporal resolution of 83 ms), 2 × 32 × 0.6 mm detector collimation with a z-flying focal spot, 0.75-mm slice width, 0.4-mm reconstruction interval, and a cycle time of 1.36 seconds for one acquisition and the subsequent table feed resulting in data acquisitions every 2 or 3 R-R intervals depending on heart rates. Sequential scanning was acquired at the end-systole of the R-R interval by using an absolute trigger delay (ms). For combined ECG and respiratory triggering, a pressure-sensing belt of a respiratory gating system (AZ-733V, Anzai Medical Co., Tokyo, Japan) was placed, usually in the upper abdomen, to obtain the respiratory signal (Fig. 1). Then, respiratory triggering was produced at either inspiration (n = 5) or expiration peak (n = 10) of the respiratory cycle determined by a pediatric radiologist based on the shape of respiratory curve in each child. Sequential scanning was acquired only when ECG triggering was synchronous with respiratory triggering through a prototype box that was not commercially available (Fig. 1). Average sequential scan rates per minute during free-breathing lung CT were recorded and compared between the two scan modes. Body size-adapted volume CT dose index derived from 16 cm-sized phantom of lung CT was determined individually by the cross-sectional area and the mean body density of each patient (22). To reduce image noise of unenhanced lung CT images, high tube voltage was used, e.g., 100 kVp in 14 patients, 120 kVp in one patient. Multiple sequential scans were obtained to cover the whole lung. The number of sequential scan slabs per lung CT was compared between ECG-triggered lung CT and combined ECG- and respiratory-triggered lung CT. CT dose estimates were calculated from dose length products derived from 16 cm-sized CT dose index phantom and age-, sex-, and tube voltage-specific conversion factors (23). The calculated effective dose was compared between ECG-triggered lung CT and combined ECG- and respiratory-triggered lung CT. For image interpretation, 3- to 5-mm thick axial, coronal, and sagittal CT images were reconstructed depending on scan length.

The degree of respiratory misregistration artifacts in the entire thorax including the chest wall, the lung outline, and the diaphragm was measured by using calipers on coronal and sagittal lung CT images (Fig. 2) and subsequently, semi-quantitatively graded on a 4-point scale (grade 1, maximal distance of misalignment between adjacent imaging slabs due to respiratory misregistration < 2 mm; grade 2, 2–5 mm; grade 3, 5–10 mm; and grade 4, > 10 mm).

For statistical analysis, statistical software (SPSS version 20.0, IBM, Chicago, IL, USA) was used. Variables were expressed as the mean with standard deviation. Respiratory misregistration artifact grades were compared between ECG-triggered lung CT studies with and without respiratory triggering by using the Wilcoxon signed rank test. Continuous variables were compared between the two lung CT studies by using the paired t test. A p value < 0.05 was considered as statistically significant.