This retrospective study was approved by the Institutional Review Board. Patients with acute pulmonary thromboembolism were excluded based on CT findings in this study (n = 34). We selected patients who underwent both echocardiography and DECT angiography between May 2009 and December 2013 at our hospital. Two hundred and seventy-four patients (114 men) who underwent echocardiography within two weeks also underwent DECT angiography DECT was performed for suspected pulmonary hypertension (PH). Of these, 63 patients (27 men, 36 women; mean age, 63.9 ± 15.5 years) were defined as a high sPAP group, with an estimated systolic pulmonary arterial pressure (sPAP) of ≥40 mmHg on echocardiography [5] - [7] . The remaining 211 (123 men, 88 women; mean age, 65.9 ± 15 years) patients were defined as low sPAP group (Table 1) [5] - [7] . In subjects with a high sPAP, the underlying diseases were determined based on medical records and CT findings by a radiologist (over ten years of experience) as follows: pulmonary artery stenosis (n = 12) [idiopathic PH (n = 10) and congenital shunt disease (n = 2)], left heart failure (n = 11) [mitral valve stenosis (n = 7), aortic valve stenosis (n = 3), and/or mitral valve regurgitation (n = 3)], lung disease (n = 15), and chronic thromboembolism (n = 25), using the Dana Point classification system [7] - [9] . Lung diseases included chronic obstructive disease (COPD) (n = 10) and interstitial pneumonia (n = 15) [idiopathic interstitial pneumonia (n = 10) and due to other collagen diseases (n = 3)].

In subjects with a low sPAP, the underlying diseases were also determined based on medical records and CT findings as follows: no disease (n = 167), pulmonary artery stenosis (n = 11) [idiopathic PH (n = 8) and congenital shunt disease (n = 3)], lung disease (n = 18), and chronic thromboembolism (n = 15). Lung diseases included chronic obstructive disease (COPD) (n = 10) and interstitial pneumonia (n = 8) [idiopathic interstitial pneumonia (n = 5) and due to other collagen diseases (n = 3)].

Note: sPAP means systoric pulmonary artery pressure; HU means Hounsfiled units.

In all patients, we used a dual-source CT scanner (Somatom Definition; Siemens Healthcare, Erlagen, Germany) in dual-energy mode. Firstly, precontrast CT of the chest was conducted.

High-concentration iodine-based contrast material (Omnipaque 350; Daiichi-Sankyo, Tokyo, Japan) was administered at a flow rate of 4.0 mL/s followed by a 40-mL saline chaser bolus at the same injection rate. The total volume of contrast material was adapted to the patient’s body weight at 1.35 ml/kg. In this study, CT scanning began at a fixed 14 s from the start of injection [3] [4] . To avoid streak artifacts due to highly concentrated contrast material in the subclavian vein or superior vena cava, scans were acquired in the caudocranial direction. Other scan parameters were as follows: tube voltage, 140 and 80 kVp at 30 and 210 effective mAs; attenuation-based tube current modulation; rotation time, 0.5 s; collimation, 14 × 1.2 mm; and pitch, 0.7. T CT angiography (CTA) images were reconstructed with a specific medium-soft convolution kernel (D30) without edge modification at a 1.0-mm slice thickness with a 1.0-mm increment. The image reconstruction system generated three stacks of axial images: 140-kV images, 80-kV images, and weighted-average images that derived 60% of the image density from the 140-kV images and 40% from the 80-kV images to achieve an image quality and a signal-to-noise ratio similar to those of single-energy 120-kV scans [3] [4] .

Doppler transthoracic echocardiography examination was performed by one cardiologist (over ten years of experience) using a GE Healthcare Vivid q (GE Medical Systems; Milwaakee, Wis). Estimated sPAP was obtained using the modified Bernoulli formula by measuring the peak tricuspid regurgitation velocity and estimation of central venous pressure from inferior vena cava compliance.

On the basis of three-material decomposition of soft tissue, iodine, and air, iodine distribution maps of the lungs were made on a workstation (Syngo MultiModality; Siemens Healthcare) with specific dual-energy post- processing software. Hounsfield units (HU) were automatically calculated in several patterns, including both lungs, and the left or right lungs [4] . We used the data from both lungs.

All values are expressed as the means. Statistical analysis was performed using clinical and morphological variables with the paired t-test and Mann?Whitney’s U-test for continuous variables. One-way analysis of variance (ANOVA) was used to test for significant differences in mean lung PBV values among the four causes of PH. If a significant difference was found, the Bonferroni-Dunn test was used for post-hoc analysis. Pearson correlation coefficients were used to examine correlations. Correlation coefficients of 0.4 - 1.0 were considered to indicate a correlation [10] . In all tests, p < 0.05 was considered significant (Statview J-5.0 for Windows; Spaces Concepts, Berkeley, Calif., USA).

Table 1 shows a summary of the results. For patient (whole lung)-based analysis, lung PBV values were 25.0 ± 9.6 and 29.0 ± 9.3 in high and low sPAP groups, respectively. There was a significant difference between the two groups (p = 0.003). However, no significant correlation between the lung PBV and PAP was observed (R = −0.2177, p = 0.063) in all subjects (n = 274) (Figure 1).

In the high sPAP group with lung diseases (n = 15), chronic thromboembolism (n = 25), pulmonary artery stenosis (n = 12), and left heart failure (n = 11), lung PBV values were 18.6 ± 1.6, 25.1 ± 4.5, 25.8 ± 4.5, and 32.7 ± 9.4 HU, respectively. One-way ANOVA was used to test for significant differences in mean lung PBV values among the four causes of PH. There were significant differences in quantification of the lung PBV among the

underlying diseases (p < 0.0001) (Figure 2).

Subjects with left heart failure had a significantly higher lung PBV than subjects with chronic thromboembolism (p = 0.0181) and pulmonary diseases (p = 0.0001). Subjects with pulmonary artery stenosis had a significantly higher lung PBV than subjects with pulmonary diseases (p = 0.0349). Subjects with chronic thromboembolism had significant higher lung PBV value than those in subjects with pulmonary disease (p = 0.0246).

In the high sPAP group, no significant correlation between the lung PBV and sPAP was observed (R = −0.171, p = 0.1796) (Figure 3).

In subjects with left heart failure, a positive correlation between the lung PBV and pulmonary artery pressure was observed (R = 0.721, p < 0.0001) (Figure 4(a)).

On the other hand, in subjects with lung disease (R = −0.487, p < 0.0001), chronic thromboembolism (R = −0.415, P = 0.0391), and pulmonary artery stenosis (R = −0.385, P = 0.0047), a negative correlation was noted between the lung PBV and sPAP (Figures 4(b)-(d)).

We did not analysis the subjects with low sPAP group because many subjects had no the underlying diseases.

PH is a life-threatening condition characterized by an elevated PAP and pulmonary vascular resistance. However, no methods have been available for the quantification of pulmonary parenchyma perfusion itself. The lung PBV may be a useful tool for its quantification in PH. Additionally, lung PBV, CT, and CTA images can be obtained at the same time. Therefore, other serious causes of PH, including pulmonary embolism, can be excluded at the same time. As a result, lung PBV images may reduce the radiation exposure of patients.

In this study, there were significant differences in quantification of the lung PBV among the underlying diseases in the high PAP group. Especially, subjects with left heart failure had a significantly higher lung PBV than subjects with other diseases. These results suggest that the lung PVB can be used to estimate the etiology of PH. Additionally, this method, with its automatic quantification, is rapid and simple to use.

There were several limitations of this study. Firstly, it included a limited number of patients, and all were diagnosed based on echocardiography. Therefore, the sPAP was not proved on right heart catheterization. Further studies are needed.

The second limitation was that coverage of the whole lung volume within this reduced FOV (26 cm) is not feasible. In these cases (81 of 253 cases), peripheral defects might be overlooked.

As the third limitation, CT scanning began at a fixed 14 s from the start of injection based on previous reports [3] [4] . This might underestimate the PBV in patients with suspected PH where the mean pulmonary transit time may be reduced.