Changes of Emphysema Parameters over the Respiratory Cycle During Free Breathing: Preliminary Results Using Respiratory Gated 4D-CT

ABSTRACT

The purpose of this research was to evaluate respiratory gated CT of the lung in patients with COPD for analysis of parenchymal characteristics who were potential candidates for volume reduction surgery.

Eleven patients with clinically known emphysematous disease underwent a respiratory gated, free-breathing 64-multislice-CT (Aquilion 64, Toshiba). Retrospective image reconstruction was performed similar to cardiac CT at every 10% of the respiratory loop, resulting in 10 complete volumetric datasets at 10 equidistant time points. All images were transferred onto a PC for calculation of the total lung volume, emphysema volume, emphysema index, and mean lung density.

Complete datasets could be successfully reconstructed in all patients. The mean lung volume increased from 6.9 L to 7.5 L over the respiratory cycle. Emphysema volume increased from 1.6 L to 2.0 L and emphysema index from 22.6% to 26.5% from expiration to inspiration.

In conclusion, respiratory gated chest 4D-CT allows for combined morphologic and functional image analysis, which can provide new insight into functional impairment and individual treatment planning.

KEYWORDS:

• Lung
• emphysema
• 4D-CT
• respiratory gating
• quantification

Introduction

Chronic obstructive pulmonary disease (COPD) and subsequent pulmonary emphysema is one of the leading diseases of the lung (1). So far, mainly static computed tomography (CT) images in inspiratory breath-hold are acquired during radiological work-up to assess the extent of emphysema (2–4). Only few papers elucidated the functional work-up with a paired inspiratory and expiratory breath-hold CT (5). However, gaining more detailed functional information is getting more important as therapeutic approaches like endobronchial valve implantation or lung volume reduction surgery are applied on a regular basis. Also, lung motion analysis and subsequent changes of regional lung motion in pathology is not yet visualized or understood (6). Thus, a true functional CT during the whole respiratory cycle might be of value for improved and in-detail treatment planning.

As shown in animal experiments and few patient studies, respiratory gated whole lung CT is feasible with a high accuracy for assessment and quantification of lung parenchymal changes (7–9). Even more recently, it was shown, that dynamic CT investigated in 3 patients allows for assessment of airways instability (10). 4D-CT is quite frequently used in radiation oncology for assessment of lung tumor motion and even for gated radiation treatment (11). Based on the maximum inspiration/expiration phase from 4D-CT dataset acquired for radiation planning of lung cancer, ventilation maps have been generated for estimation of post-surgical lung function (12).

So far, there are no studies out exploring this new technology in patients with COPD (without any lung cancer burden) with quantification of emphysema parameters over the respiratory cycle. Therefore, the aim of this pivotal study was to demonstrate the feasibility of 4D-CT assessment in patients with COPD and provide insights into quantitative image analysis with focus on emphysema parameters.

Material and methods

Patients

11 consecutive patients with clinically known emphysema (3 female and 8 male) with the mean age of 68 ± 5 years (range 60–78) were included in this study (mean weight 65 ± 16 kg; mean height 170 ± 8 cm). All patients had a smoking history and significant impaired lung function: FEV1% predicted 30 ± 9%; residual volume% predicted 247 ± 56%; total lung capacity% predicted 129 ± 22%; peak expiratory flow 33 ± 9%. The CT examination was performed as part of clinical work-up for patients who due^{Figure 1} to clinical impairment might be potential candidates for volume reduction therapy. The study was approved by the local ethics committee.

Visual CT evaluation reveals 4 patients with homogeneous distribution of emphysema and 7 with heterogeneous distribution. Assuming 3 lobes per lung resulting in 66 lobes, there were 8 lobes with predominately destructive emphysema, 31 with confluent centrilobular emphysema and 27 with intermediate centrilobular emphysema type.

CT scan protocol

CT was performed using a respiratory gated non-enhanced 64-multislice CT (Aquilion-64, Toshiba, Japan) during normal free breathing in supine position. Every patient was carefully instructed how to breathe before the scan. CT parameters were: collimation 1 mm, 120 kV, 50 mA, half-scan reconstruction with rotation time 0.4 seconds (effective gantry rotation time 0.2 seconds), reconstructed slice thickness 1 mm and increment 0.8 mm. No dose modulation was used. A mean helical pitch of 7.8/64 (0.12) was used. Retrospective image reconstruction was performed similar to cardiac CT at every 10% of the respiratory cycle, resulting in 10 complete volumetric datasets of the whole lung at 10 equidistant time points. All images were reconstructed using a soft tissue kernel (FC02). An image example illustrating the image quality is shown in Figure 1.

Figure 1. Coronal images at maximum expiratory level (A, phase 20%) and maximum inspiratory level (B, phase 60%) at the identical anatomical position showing the diagnostic quality of the data.

Gating device

A flow measuring sensor was put into patients nose, the sensor was connected to a device (Breas Medical). This device was able to visualize flow curves and also provided a digital output. This output signal was connected to the scanner instead of the ECG gating unit (13). The CT acquisition was then gated to this signal, comparable to a cardiac CT examination, only with respiration as input.

Image analysis

Semiautomatic lung segmentation was done using dedicated software for emphysema analyses (YACTA®, Germany), running on a standard PC-platform. Important morphological thoracic landmarks i.e. trachea, right and left lung were automatically detected. Trachea and bronchi up to the eighth generation were excluded from the evaluation. On the basis of the pulmonary landmarks, the lung was automatically detected by a region growing with a N6 neighborhood system and an upper threshold of −500 HU (this resulted in a “safe” segmentation of the lung parenchyma without surrounding thoracic structures). All lung voxels below −950 HU were segmented as emphysema. This was followed by a correction factor which included all voxels from −950 to −910 HU if they were surrounded by emphysema voxels. From this analysis we received the total lung volume (LV), emphysema volume (EV), emphysema index (EI), and mean lung density (MLD) for the whole lung for each step of respiratory cycle (Figure 2).

Figure 2. Changes of lung volume (A), emphysema volume (B) and emphysema index (C) over the whole respiratory cycle for each patient. One respiratory cycle (0–90%) was acquired and data were duplicated (100–190%) for better graphical visualization of the respiratory changes.

Results

All datasets were eligible for analysis with good image quality. For every patient, 10 datasets were reconstructed; so overall 110 CT datasets were analysed.

The results for each patient were presented in Table 1. The mean maximum MLD was −854 ± 13 HU L, and the mean minimum MLD was −865 ± 10 HU. The mean maximum LV was 7.5 ± 1.4 L, and the mean minimum LV was 6.9 ± 1.4 L. The mean maximum EV was 2 ± 0.7 L, and the mean minimum EV was 1.6 ± 0.7 L. The mean maximum EI was 26.5 ± 7.7%, and the mean minimum EI was 22.6 ± 8.2%.

Table 1. Individual results for the range and mean () of lung volumes, mean lung densities, emphysema volumes and emphysema indices over the respiratory cycle.

Patient	LV [L]	MLD [HU]	EV [L]	EI [%]
1	6.26–6.86 (6.53)	−867.7 to −859.2 (−863)	2.23–2.55 (2.38)	35.1–38.0 (36.4)
	Δ 0.6	Δ 8.6	Δ 0.33	Δ 2.9
2	4.17–4.99 (4.55)	−858.0 to −837.7 (−846)	0.73–1.15 (0.92)	17.4–23.1 (20.1)
	Δ 0.82	Δ 20.3	Δ 0.42	Δ 5.7
3	7.81–8.43 (8.1)	−869.0 to −862.0 (−865)	1.77–2.14 (1.95)	22.7–26.4 (24.1)
	Δ 0.61	Δ 7	Δ 0.37	Δ 3.7
4	9.12–9.90 (9.45)	−871.4 to −861.9 (−865)	2.39–2.82 (2.56)	25.6–28.9 (27.1)
	Δ 0.78	Δ 9.5	Δ 0.44	Δ 3.3
5	6.48–7.15 (6.77)	−852.6 to −840.6 (−845)	0.70–0.98 (0.83)	10.8–14.1 (12.2)
	Δ 0.67	Δ 12	Δ 0.28	Δ 3.3
6	6.67–7.17 (6.9)	−880.0 to −874.5 (−877)	2.36–2.73 (2.54)	35.4–38.1 (36.9)
	Δ 0.50	Δ 5.5	Δ 0.37	Δ 2.7
7	8.27–9.00 (8.6)	−878.1 to −865.0 (−872)	1.58–2.10 (1.87)	19.1–23.3 (21.7)
	Δ 0.73	Δ 13.1	Δ 0.52	Δ 4.3
8	7.24–7.78 (7.52)	−874.7 to −868.2 (−871)	2.15–2.55 (2.37)	29.6–34.2 (31.5)
	Δ 0.54	Δ 6.5	Δ 0.4	Δ 4.6
9	8.06–8.86 (8.42)	−855.1 to −839.3 (−848)	1.02–1.59 (1.23)	12.4–18.5 (14.6)
	Δ 0.81	Δ 15.8	Δ 0.57	Δ 6.1
10	5.65–6.16 (5.87)	−861.2 to −849.2 (−855)	1.14–1.44 (1.28)	20.2–23.4 (21.8)
	Δ 0.51	Δ 11.9	Δ 0.3	Δ 3.2
11	5.71–6.36 (6.01)	−853.6 to −840.4 (−846)	1.14–1.45 (1.28)	20.0–22.8 (21.3)
	Δ 0.64	Δ 13.1	Δ 0.31	Δ 2.8

The individual change of LV over the respiratory cycle ranged between 0.5 L and 0.82 L (mean 0.66 L, equals 9%). The individual change of EV ranged between 0.28 L and 0.57 L (mean 0.39 L, equals 22%) and of EI ranged between 2.7% and 6.1% (mean 3.9%, equals 16%). The changes over the respiratory cycle are presented in Figure 2.

Discussion

This is the first study using 4D-CT in patients suffering from COPD with detailed analysis of emphysema changes throughout the whole respiratory cycle. The main findings of this feasibility study were that quantification of lung volumes and emphysematous changes over the respiratory cycle can be acquired and allow for quantification (Figures 2 and 3). Our study allowed for detailed whole-lung analysis of respiratory parameters in non-tumor patients providing new insights without any comparison in the literature.

Figure 3. Segmentation results and visualization of the whole lung and the emphysematous areas (in yellow) during the whole respiratory cycle (from 0% to 90%). The lung volume in detail was: 0% – 5889 ml, 10% – 5777 ml, 20% – 5714 ml, 30% – 5754 ml, 40% – 5980 ml, 50% – 6251 ml, 60% – 6359 ml, 70% – 6313 ml, 80% – 6079 ml, 90% – 5963 ml. The emphysema index in detail was: 0% – 20.2%, 10% – 20.1%, 20% – 20.0%, 30% – 20.2%, 40% – 21.1%, 50% – 22.3%, 60% – 22.8%, 70% – 22.6%, 80% – 22.1%, 90% – 21.5%.

In our study we found the mean change of EI during respiration of 16%. While this change was measured during free-breathing, our data is similar to those found in previous studies analyzing paired CT scans at deep inspiration and deep expiration (mean inspiratory EI 18% and change 11% (14), mean inspiratory EI 27% and change 5% (15), mean inspiratory EI 21% and change 9% (16), mean inspiratory EI 14% and change 5% (17), mean inspiratory EI 21% and change 8% (18), mean inspiratory EI 52% and change 11% (19)).

As mentioned above, our data were acquired during normal respiration. Thus, the extreme values like in full-inspiration and full-expiration breath-hold will never be reached and more physiological values will be achieved. As a positive effect, there is no glottis closure, thus no valsalva maneuver is performed during the examination (otherwise like in inspiratory breath-hold), which might introduce a negative effect on the airways and therefore lung inflation.

Furthermore, it has to be kept in mind that the patients included in this study suffered from severe emphysema. Therefore, the data presented are only true for severely destructed lungs and not in a normal population or mild stage of the disease.

During the breathing pattern of the patients a mean change between expiration and inspiration of LV of 0.66L and an increase of EV of 0.39L was seen. This means that most of the tidal volume resulted in a hyperinflation of lung parenchyma. This finding was previously described in a study in inspiration and expiration CT in emphysematous patients (19). This can be explained by the lung areas with functional hyperinflation, where the lung density values at inspiration are within the emphysema threshold for segmentation purposes and increase during expiration and therefore are not captured as emphysema anymore. This explains the pronounced change in EV during the respiratory cycle.

These parameters allow for further risk and treatment stratification as the individual change of EV may predict the functional outcome after endobronchial treatment like valve placement.

As a drawback of the technique itself, it has to be mentioned that the acquisition of CT datasets requires a regular breathing pattern. Like in cardiac gated CT, where the cardiac frequency has to be kept as low and regular as possible to get the best images, for respiratory gated CT acquisition these parameters have to be optimized as well keeping in mind the temporal resolution of 250 msec (7). Therefore, no information about forced respiratory maneuvers can be acquired and evaluated.

Increase of noise by using low dose images has an impact on the quantitative evaluation and can distort the results (20). Therefore no dose modulation techniques were applied as the image noise should be kept constant through the respiration phases not to influence the quantitative results.

Analyzing the graphs in more detail, it is noticed that the LV changes most in the first third of the respiratory cycle, while the EI increases quite constantly over the respiratory cycle. Using respiratory gating and a wide collimation CT scanner (320 slices) 4D-CT data can be acquired covering only 16 cm of the lung. This technique was used in a recently published paper, where the main MLD change was also observed within the first part of the respiratory cycle (21). This information may be used to focus on the crucial part and use less radiation exposure in less relevant part of the respiratory cycle, according to cardiac ECG gating where a phase range can be selected to reduce radiation dose. Early literature on 4D-CT discussed that image acquisition increase patient dose compared to a single breath-hold acquisition, which depending on scan parameters can increase by up to an order of magnitude (22). Therefore, we used a reduced dose protocol for our data acquisition (50mA with 0.4 seconds rotation time = 20 mAs). However, keeping in mind that patients with COPD might get improved quality of life with tailored and individual therapy, which might be influenced by in-depth analysis of the whole respiratory cycle, this would outweigh the radiation issue.

Regarding individualized therapy, it has to be mentioned that endobronchial treatment is already planned on the individual disease distribution, based on one inspiratory with or without additional expiratory CT dataset. This feasibility study could show a high data quality for whole lung segmentation and quantitative evaluation during the whole respiratory cycle. This kind of information can be used to enhance and improve further evaluation of regional lung function like for vector analysis of lung motion, so far investigated on paired inspiratory/expiratory CT (6). Also, regional ventilation maps could be generated in more detail based on a steady assessment and tracking of the lung parenchyma throughout the respiratory cycle. So far the ventilation maps were based on inter- and extrapolation from either paired inspiratory/expiratory CT or maximum inspiration/expiration phase from 4D-CT (12, 23). Therefore, the so far published CT derived ventilation maps are based on huge assumptions what happens between the two extreme points.

After this feasibility study, further studies can be performed with focus on a regional, i.e. lobar, analysis of the ventilation parameters.

In conclusion, we could demonstrate that dynamic 4D-CT datasets can be acquired in patients with COPD allowing for functional insights into the dynamics of temporal kinetics of the changes.

Declaration of interest statement

All authors have no financial interest that could influence (bias) the author's work.