**4.1 Image post-processing**

From the raw spiral projection data of both tubes, images were automatically reconstructed to three separate image sets: 80 kVp, 140 kVp and average weighted virtual 120 kVp images with 80:140 kVp linear weighting of 0.3 (i.e. 30% image information from the 80 kVp image and 70% information from the 140 kVp image). For each image set, the slice thickness was 0.75 mm and interval was 0.50 mm. The only currently commercially available software (syngo DE Lung PBV by Siemens HealthCare) for DECT lung perfusion image analysis is part of the dual energy post-processing software package available for the Siemens syngo MultiModality Workplace. For the calculation of iodine distribution in the lung parenchyma, the application class is designed for iodine extraction, and the material parameters for iodine extraction are as follows: -1,000 HU for air at 80 kVp, -1,000 Hounsfield unit (HU) for air at 140 kVp, 60 HU for soft tissue at 80 kVp, 54 HU for soft tissue at 140 kVp, 2 for relative contrast enhancement, -960 HU for minimum value, -200~- 300 HU for maximum value, and 4 for range (Figure 1A).The lung parenchyma is color coded using gray scale 16-bit or hot metal 16-bit color coding (default setting) with different optional color scales available. The software enables a multiplanar view of the lung parenchyma image. The software also enables users to set a mixing ratio between a non color-coded virtual 120 kV dataset and color-coded lung parenchyma. This mixing ratio can be fluently set between 0% showing a anatomy image and 100% showing a blood flow image (BFI) images where only the color-coded, segmented lung parenchyma is displayed. Windowing functionality for the original and color-coded dataset, basic measurement tools, and a few dual-energy-specific measurements are also available. The fused images are obtained by mixing the anatomy image and BFI images with different ratios. The fused images are used for visualization of CT pulmonary angiography and the lung perfusion.

Fig. 1. Parameter settings of the postprocessing software for dual energy CT pulmonary angiography

A) Parameter settings of Lung Pulmonary Blood Volume (PBV) software; B) Parameter settings of Lung Vessels software

The Lung Vessels application was developed to discriminate non-enhancing subsegmental pulmonary arteries from enhancing ones by using dual energy iodine extraction data. This technique had a high negative predictive value being important for exclusion of segmental PE. In the Lung Vessels application, results are displayed as color-coded multi planar reformatted data and a 3D volume rendered dataset, where vessels with high iodine content are colorcoded blue and soft tissue or vessels with low or no iodine content due to PE are color-coded red. The material parameters for iodine extraction of Lung Vessels are as follows: -1,000 HU for air at 80 kVp, -1,000 Hounsfield unit (HU) for air at 140 kVp, 60 HU for soft tissue at 80 kVp, 54 HU for soft tissue at 140 kVp, 1.1 for relative contrast enhancement, -500 HU for minimum value, 3071 HU for maximum value, and 4 for range (Figure 1B).

#### **4.2 BFI image interpretation**

208 Pulmonary Embolism

From the raw spiral projection data of both tubes, images were automatically reconstructed to three separate image sets: 80 kVp, 140 kVp and average weighted virtual 120 kVp images with 80:140 kVp linear weighting of 0.3 (i.e. 30% image information from the 80 kVp image and 70% information from the 140 kVp image). For each image set, the slice thickness was 0.75 mm and interval was 0.50 mm. The only currently commercially available software (syngo DE Lung PBV by Siemens HealthCare) for DECT lung perfusion image analysis is part of the dual energy post-processing software package available for the Siemens syngo MultiModality Workplace. For the calculation of iodine distribution in the lung parenchyma, the application class is designed for iodine extraction, and the material parameters for iodine extraction are as follows: -1,000 HU for air at 80 kVp, -1,000 Hounsfield unit (HU) for air at 140 kVp, 60 HU for soft tissue at 80 kVp, 54 HU for soft tissue at 140 kVp, 2 for relative contrast enhancement, -960 HU for minimum value, -200~- 300 HU for maximum value, and 4 for range (Figure 1A).The lung parenchyma is color coded using gray scale 16-bit or hot metal 16-bit color coding (default setting) with different optional color scales available. The software enables a multiplanar view of the lung parenchyma image. The software also enables users to set a mixing ratio between a non color-coded virtual 120 kV dataset and color-coded lung parenchyma. This mixing ratio can be fluently set between 0% showing a anatomy image and 100% showing a blood flow image (BFI) images where only the color-coded, segmented lung parenchyma is displayed. Windowing functionality for the original and color-coded dataset, basic measurement tools, and a few dual-energy-specific measurements are also available. The fused images are obtained by mixing the anatomy image and BFI images with different ratios. The fused images are used for visualization of CT pulmonary angiography and the

**4. Image post-processing and image interpretation** 

A B

Fig. 1. Parameter settings of the postprocessing software for dual energy CT pulmonary

A) Parameter settings of Lung Pulmonary Blood Volume (PBV) software; B) Parameter

**4.1 Image post-processing** 

lung perfusion.

angiography

settings of Lung Vessels software

It is very important to recognize the normal findings or artifacts at DECT lung perfusion. Normal pulmonary BFI images were defined as showing homogeneous perfusion in the normal range (color-coded yellow-green or blue) with dependent symmetric lung iodine distribution (Figure 2). Dependent lung perfusion at DECT refers to relatively low contrast enhancement in the ventral regions (color coded yellow-green) and relatively higher enhancement in the dorsal regions (color coded blue -black) with the patient in the supine position (Figure 3).

Fig. 2. Normal pulmonary blood flow imaging

A) Axial BFI image and B) coronal fused image show homogeneous blood flow distribution in both lungs

In the analysis of BFI images, sources of pitfall should be kept in mind to avoid misdiagnoses. When interpreting BFI images, these pitfalls can relate to artifacts from contrast material, diaphragmatic or cardiac motion, pulmonary pathology and the occlusive degree of pulmonary arteries.

Streak and beam-hardening effects resulting from high-concentration contrast agent in the thoracic veins and right cardiac chambers can commonly cause heterogeneous artifacts in BFI images (Figure 4); these artifacts must be considered when an unexpected contrast enhancement defect is noted adjacent to an area of high contrast enhancement. In this setting, the perfusion defect may appear band-like and be mostly in both upper lobes. Optimization of contrast medium injection parameters, including the use of a saline chaser, can reduce the beam-hardening artifact, improve the image quality of DECT and increase

Dual Source, Dual Energy Computed Tomography in Pulmonary Embolism 211

Crescent-shaped perfusion defect is seen in the lung parenchyma adjacent to the cardiac

 A B Fig. 6. Contrast defect in the BFI image caused by emphysema

In addition, the anatomy CT images should be evaluated for pulmonary pathology, such as emphysema (Figure 6), tumors invading or compressing the pulmonary arteries (Figure 7) and pulmonary consolidation (Figure 8), all of which will result in contrast enhancement defects in BFI images. The occlusive degree of pulmonary arteries will affect perfusion defects at BFI and result in the false-negatives (Figure 9). But, misdiagnosis resulting from these factors is rare when BFI images are interpreted in conjunction with CTPA, which can reliably detect the lobar and segmental emboli. Nevertheless, small peripheral pulmonary emboli causing minimal contrast enhancement defect alterations can be overlooked even when state-of-the-art MDCT scanners are employed. Also, the considerable reduction in the pulmonary capillary bed often seen in the elderly or patients with emphysema can cause diffuse decreased pulmonary ''perfusion''

A) A coronal BFI image shows heterogeneous contrast enhancement in both lungs caused by the emphysema that is readily seen at coronal reformatted multiplanar reformation viewed

Fig. 5. Cardiac motion artifact

chambers (white arrow)

**[Boroto K, 2008].** 

with lung windows (B)

diagnostic confidence. Nance JW Jr et al [,Nance JW Jr,2011] reported that iomeprol 400 at 4 mL/s (an IDR of 1.6 g I/s) resulted in superior quality CTPA and perfusion map images compared with the protocols using a lower concentration or delivery rate.

Fig. 3. Gravity-dependent lung perfusion states A normal pulmonary BFI image obtained in one patient in the supine position shows relatively low pulmonary contrast enhancement anteriorly (arrows) and relatively high

contrast enhancement in more dependent lung portions

Diaphragmatic or cardiac motion can cause apparent lower areas of lung contrast enhancement in the lung parenchyma adjacent to the diaphragm or cardiac chambers (Figure 5). In this setting, the perfusion defect is crescent-shaped; blurring or double lines adjacent to the diaphragm or heart border can be seen on images obtained with lung windows or mediastinal windows. Patients should hold their breath while scanning to reduce the diaphragm motion artifacts. A potential method to improve image quality in the vicinity of the cardiac chambers might be to synchronize data acquisition with electrocardiographic tracing, a technological development currently exclusively available for myocardial ''perfusion'' analysis **[Pontana F,2008]**; however, such methodology could also potentially introduce stair-step or misregistration artifacts and requires further study. Normal physiological gravity dependent variation in pulmonary ''perfusion'' should also be recognized **[Zhang LJ,2009(Eur Rdiol)/2009(Acta Radio)].**

Fig. 4. Pseudo-high perfusion due to dense contrast material in the superior vena cava A) An axial BFI image shows radiating pseudo-high perfusion and pseudo-iodine defect adjacent to the superior vena cava (white arrow) due to streak artifact from high concentration contrast material. B) Coronal maximum intensity projection image shows a higher opacity of superior vena cava (red arrow) than pulmonary artery

#### Fig. 5. Cardiac motion artifact

210 Pulmonary Embolism

diagnostic confidence. Nance JW Jr et al [,Nance JW Jr,2011] reported that iomeprol 400 at 4 mL/s (an IDR of 1.6 g I/s) resulted in superior quality CTPA and perfusion map images

A normal pulmonary BFI image obtained in one patient in the supine position shows relatively low pulmonary contrast enhancement anteriorly (arrows) and relatively high

Diaphragmatic or cardiac motion can cause apparent lower areas of lung contrast enhancement in the lung parenchyma adjacent to the diaphragm or cardiac chambers (Figure 5). In this setting, the perfusion defect is crescent-shaped; blurring or double lines adjacent to the diaphragm or heart border can be seen on images obtained with lung windows or mediastinal windows. Patients should hold their breath while scanning to reduce the diaphragm motion artifacts. A potential method to improve image quality in the vicinity of the cardiac chambers might be to synchronize data acquisition with electrocardiographic tracing, a technological development currently exclusively available for myocardial ''perfusion'' analysis **[Pontana F,2008]**; however, such methodology could also potentially introduce stair-step or misregistration artifacts and requires further study. Normal physiological gravity dependent variation in pulmonary ''perfusion'' should also be

compared with the protocols using a lower concentration or delivery rate.

Fig. 3. Gravity-dependent lung perfusion states

contrast enhancement in more dependent lung portions

recognized **[Zhang LJ,2009(Eur Rdiol)/2009(Acta Radio)].**

A B

Fig. 4. Pseudo-high perfusion due to dense contrast material in the superior vena cava A) An axial BFI image shows radiating pseudo-high perfusion and pseudo-iodine defect

concentration contrast material. B) Coronal maximum intensity projection image shows a

adjacent to the superior vena cava (white arrow) due to streak artifact from high

higher opacity of superior vena cava (red arrow) than pulmonary artery

Crescent-shaped perfusion defect is seen in the lung parenchyma adjacent to the cardiac chambers (white arrow)

In addition, the anatomy CT images should be evaluated for pulmonary pathology, such as emphysema (Figure 6), tumors invading or compressing the pulmonary arteries (Figure 7) and pulmonary consolidation (Figure 8), all of which will result in contrast enhancement defects in BFI images. The occlusive degree of pulmonary arteries will affect perfusion defects at BFI and result in the false-negatives (Figure 9). But, misdiagnosis resulting from these factors is rare when BFI images are interpreted in conjunction with CTPA, which can reliably detect the lobar and segmental emboli. Nevertheless, small peripheral pulmonary emboli causing minimal contrast enhancement defect alterations can be overlooked even when state-of-the-art MDCT scanners are employed. Also, the considerable reduction in the pulmonary capillary bed often seen in the elderly or patients with emphysema can cause diffuse decreased pulmonary ''perfusion'' **[Boroto K, 2008].** 

Fig. 6. Contrast defect in the BFI image caused by emphysema A) A coronal BFI image shows heterogeneous contrast enhancement in both lungs caused by the emphysema that is readily seen at coronal reformatted multiplanar reformation viewed with lung windows (B)

Dual Source, Dual Energy Computed Tomography in Pulmonary Embolism 213

A B C

A B C

A) Coronal and B) Axial BFI images show a wedge-shaped perfusion defect in the left lung lower lobe dorsal segment (white circle). Pseudo-high contrast enhancement is seen in the anterior portion of the right middle lung anterior to the normal pulmonary contrast enhancement seen in the right middle lobe more posteriorly (arrows). C) Axial contrastenhanced CT image shows a corresponding occlusive filling defect representing pulmonary emboli in the left lower lobe segmental pulmonary arteries (arrow), and non-occlusive

Several studies have examined DECT for the detection of PE. Fink et al **[Fink C, 2008]**  reported that both sensitivity and specificity of DECT for the assessment of PE were 100% on a per patient basis. On a per segment basis, the sensitivity and specificity ranged from 60%–66.7% and from 99.5%–99.8%; CTPA was used in this study as the standard of reference in 24 patients with suspected PE, 4 of whom actually had PE. With scintigraphy as the standard of reference, Thieme et al **[ Thieme SF ,2008]** reported 75% sensitivity and 80% specificity on a per patient basis and 83% sensitivity and 99% specificity on a per segment basis in a small group of patients with DECT. A group of 117 patients was examined by Pontana et al **[Pontana F, 2008]** to investigate the accuracy of DECT in the depiction of perfusion defects in patients with acute PE, concluding that simultaneous information on the presence of endoluminal thrombus and lung perfusion impairment can be obtained with

PE(white arrow) and result in the false-negatives

Fig. 10. Acute PE in a 24-year-old man

emboli elsewhere

Fig. 9. Negative BFI image in one patient with left lower pulmonary artery embolus A) axial, B) coronal, and C) sagittal fused images show normal findings with non-occlusive

Fig. 7. Contrast defect in the pulmonary blood volume image caused by lung carcinoma A) A coronal MIP image shows a left lung hilar carcinoma invading the left pulmonary lobar arteries (red arrow), resulting in diffuse decreased contrast enhancement of the left lung ( red circle )at the corresponding coronal BFI image fused with the CT angiogram (B)

Fig. 8. Contrast enhancement defect in the pulmonary blood volume image caused by lung consolidation

A) Axial BFI image shows a contrast enhancement defect in the left lower lobe (white circle); B) The corresponding axial CT image clearly shows pulmonary consolidation in the corresponding left lower lung lobe (white circle)
