Vous êtes sur la page 1sur 9

European Journal of Radiology 87 (2017) 3644

Contents lists available at ScienceDirect

European Journal of Radiology

journal homepage: www.elsevier.com/locate/ejrad

Original research

Parietal pleural invasion/adhesion of subpleural lung cancer:

Quantitative 4-dimensional CT analysis using dynamic-ventilatory
Kotaro Sakuma a,b , Tsuneo Yamashiro c, , Hiroshi Moriya a , Sadayuki Murayama c ,
Hiroshi Ito b
Department of Radiology, Ohara General Hospital, 6-11 Omachi, Fukushima City, Fukushima 960-8611, Japan
Department of Radiology and Nuclear Medicine, Fukushima Medical University, 1 Hikariga-oka, Fukushima City, Fukushima 960-1295, Japan
Department of Radiology, University of the Ryukyus, 207 Uehara, Nishihara, Okinawa 903-0215, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Purpose: Using 4-dimensional dynamic-ventilatory scanning by a 320-row computed tomography (CT)
Received 26 August 2016 scanner, we performed a quantitative assessment of parietal pleural invasion and adhesion by peripheral
Received in revised form 1 December 2016 (subpleural) lung cancers.
Accepted 4 December 2016
Methods: Sixteen patients with subpleural lung cancer underwent dynamic-ventilation CT during free
breathing. Neither parietal pleural invasion nor adhesion was subsequently conrmed by surgery in 10
patients, whereas the other 6 patients were judged to have parietal pleural invasion or adhesion. Using
Computed tomography
research software, we tracked the movements of the cancer and of an adjacent structure such as the rib
Lung cancer
Pleural adhesion
or aorta, and converted the data to 3-dimensional loci. The following quantitative indices were compared
Four-dimensional scanning by the Mann-Whitney test: cross-correlation coefcient between time curves for the distances moved
Ventilation from the inspiratory frame by the cancer and the adjacent structure, the ratio of the total movement
distances (cancer/adjacent structure), and the cosine similarities between the inspiratory and expiratory
vectors (from the cancer to the adjacent structure) and between vectors of the cancer and of the adjacent
structure (from inspiratory to expiratory frames).
Results: Generally, the movements of the loci of the lung cancer and the adjacent structure were similar
in patients with parietal pleural invasion/adhesion, while they were independent in patients without.
There were signicant differences in all the parameters between the two patient groups (cross-correlation
coefcient and the movement distance ratio, P < 0.01; cosine similarities, P < 0.05).
Conclusion: These observations suggest that quantitative indices by dynamic-ventilation CT can be utilized
as a novel imaging approach for the preoperative assessment of parietal pleural invasion/adhesion.
2016 Elsevier Ireland Ltd. All rights reserved.

1. Introduction multidetector CT (MDCT) scanner (Aquilion ONE by Toshiba Med-

ical Systems) and 256-row MDCT scanner (Revolution CT by GE
Dynamic-ventilation computed tomography (CT) is a novel uo- Healthcare). With the use of powerful iterative reconstruction
roscopic imaging method for the thorax, which continuously scans techniques, current dynamic-ventilation CT covers longer respi-
the thorax under free breathing conditions (maximum 160 mm in ratory cycles with limited radiation exposure. The advantages of
length). Currently, this scanning method can be provided by com- 4-dimensional dynamic-ventilation CT for observing airway or
mercially available CT scanners that have a very long scanning pulmonary disorders have gradually been revealed in reports on
length in the cranio-caudal axis (z-axis), which are the 320-row such disorders as central airway stenosis in children and chronic
obstructive pulmonary disease [14].
Although a limited number of publications are available,
dynamic-ventilation CT has been applied to the preoperative anal-
Corresponding author at: Department of Radiology, Graduate School of Medical
ysis of lung cancer and mediastinal liposarcoma [57]. The main
Science, University of the Ryukyus, 207 Uehara, Nishihara, Okinawa 903-0215, Japan.
E-mail addresses: ksakuma@ohara-hp.or.jp (K. Sakuma), clatsune@yahoo.co.jp
purpose of preoperative dynamic-ventilation CT in these publi-
(T. Yamashiro), hrshmoriya@gmail.com (H. Moriya), sadayuki@med.u-ryukyu.ac.jp cations was to determine tumor resectability based on invasion
(S. Murayama), h-ito@fmu.ac.jp (H. Ito).

0720-048X/ 2016 Elsevier Ireland Ltd. All rights reserved.
K. Sakuma et al. / European Journal of Radiology 87 (2017) 3644 37

to the parietal pleura or surrounding structures such as the aorta gists or thoracic surgeons and arranged by radiologists. Pleural
or diaphragm. Imaging evidence of the movement of a peripheral invasion or adhesion was not subsequently conrmed by surgery in
(subpleural) lung cancer independent of the movement of the pari- 10 patients. The remaining 6 patients were judged to have pleural
etal or mediastinal pleura indicates that the cancer does not invade invasion or adhesion by the subsequent surgical procedure (n = 2)
or strongly adhere to the pleura, and therefore is usually resectable or by observations on the dynamic-ventilation CT (n = 4).
[7]. Even when radical surgery is selected for advanced lung cancer The lung cancer was located in the right upper lobe (n = 4), right
with invasion to the parietal pleura or mediastinum, an accurate middle lobe (n = 1), right lower lobe (n = 4), left upper lobe (n = 4),
preoperative assessment of tumor invasion or adhesion can lead and left lower lobe (n = 3). The maximum tumor diameter ranged
to appropriate informed consent before the procedure, and allow from 13 to 60 mm (mean 28 mm) on axial CT imaging. Histopatho-
the physician to prepare for longer operating time or greater blood logical examinations revealed adenocarcinoma (n = 10), squamous
loss. Currently, only qualitative (visual) assessment is used to evalu- cell carcinoma (n = 3), large cell neuroendocrine carcinoma (n = 1),
ate tumor invasion of the parietal pleura or surrounding structures and metastatic lung cancer from resected rectal cancer (n = 1). One
[57], and, to the best of our knowledge, there are no available patient who did not receive radical surgery underwent broncho-
published reports on investigations of quantitative indices of the scopic biopsy, but his lung tumor was diagnosed as indeterminate
movements of tumor and adjacent structures, as determined by malignancy due to a small specimen size.
dynamic-ventilation CT. Qualitative analysis performed by experi-
enced physicians can be replaced by quantitative assessment, and 2.2. Dynamic-ventilation CT
may be further developed to include automatic assessment by spe-
cialized software or workstations. All patients were scanned on a 320-row MDCT scanner (Aquil-
Although lung cancer invasion of the chest wall or aorta has ion ONE, Toshiba Medical Systems, Otawara, Tochigi, Japan). Using
been investigated by studies using cine-magnetic resonance (MR) the wide volume mode (non-helical mode), dynamic scanning was
imaging or combined inspiratory and expiratory chest CT [812], performed at a xed point without bed movement, resulting in a CT
these methods do not yet appear to be mainstream diagnostic uoroscopy of 160 mm in length. The scanning eld of view (FOV)
approaches and are still considered to be experimental. Consider- was selected from two settings, based on the patients body habitus:
ing that preoperative MR is an unusual imaging procedure for lung 320 mm (medium, n = 8) or 400 mm (large, n = 8). Other scanning
cancer, and expiratory chest CT requires very careful interpreta- and reconstruction parameters for the dynamic-ventilatory scan
tion by experienced chest radiologists, there remains the need for were as follows: tube current = 40 mA, tube voltage = 120 kVp,
a new, convenient imaging technique such as dynamic-ventilation rotation time = 0.35 s, total scanning time = 4.56.5 s, imaging FOV:
CT, for the preoperative assessment of tumor invasion. 320366 mm, collimation and slice thickness = 0.5 mm, reconstruc-
Based on previous publications reporting that visual assessment tion kernel = FC52 (for lung), reconstruction interval = 0.5 s/frame
performed on dynamic-ventilation CT scans was able to obtain cor- (total 913 frames), reconstruction method = half reconstruc-
rect diagnoses of tumor invasion [57], it can be hypothesized that tion. Scan data were converted to CT images using an iterative
these qualitative assessments could be converted to quantitative reconstruction method (adaptive iterative dose reduction using
evaluations by application of selected motion parameters. In this three-dimensional processing [AIDR3D], mild setting). Before
preliminary study, as the rst investigators to examine the use of the dynamic-ventilation scan, patients were coached to perform
quantitative analysis of dynamic-ventilation CT, we aimed (i) to repeated deep breathing. Radiologic technologists monitored the
identify the quantitative CT parameters that can be used to assess patients respiratory movements and started the scan just before
parietal pleural invasion/adhesion caused by peripheral (subpleu- peak inspiration. At least one cycle from the inspiratory to the
ral) lung cancer, and (ii) to examine whether these quantitative expiratory phase was included in the scan.
parameters can be used to distinguish patients with and without
parietal pleural invasion/adhesion. 2.3. Image analysis

All scan data were anonymized and stored in a research

2. Materials and methods
computer. Special in-house software that was based on com-
mercially available software (4D Airways Analysis, Toshiba
This retrospective study was approved by the Institutional
Medical Systems) synthesized the data of continuous images (320
Review Board of Ohara General Hospital, and written informed con-
images/frame total 9 13 frames) for the cine-CT series. Detailed
sent from enrolled patients was waived. This study was designed as
information on the basic software was described in a previous
a component of the ACTIve Study, an ongoing multicenter research
technical note [13]. The software was originally developed to
project in Japan. Based on agreement with the research committee
track a single designated airway point throughout the dynamic-
of the ACTIve Study Group, this study was planned as a prelimi-
ventilation CT scan. This process is facilitated by locating all
nary investigation of the potential of dynamic-ventilation CT using
anatomical positions in a single frame and automatically transfer-
320-row MDCT scanners.
ring the same position to all the other frames. In this study, we
expanded the software function that registered all spatial informa-
2.1. Subjects tion and tracked designated anatomical points, in order to follow
designated points in the lung cancer and adjacent structures and to
From October 2013 to November 2015, 16 patients with a record their foci throughout the dynamic CT scan.
peripheral lung cancer that was found to be located in the sub- A radiologist (K.S., 8 years of experience in thoracic radiology)
pleural area of the lung (within 10 mm from the pleura) underwent set two points for measurements on the inspiratory frame of the
dynamic-ventilation CT at Ohara Medical Center, which is attached dynamic-ventilation CT. One point was placed in the center of the
to Ohara General Hospital. There were 6 females and 10 males targeted lung cancer. The other point was placed in the extrapul-
(median age 75 years, range 4380 years). The dynamic-ventilation monary structure adjacent to the cancer; in most patients (n = 13),
scans were originally planned to assess the resectability of the can- the extrapulmonary measuring point was placed in the rib nearest
cer, based on cancer invasion or adhesion to the parietal pleura and the lung cancer, since these cancers had been observed to be located
adjacent (surrounding) structures such as the aorta, diaphragm, immediately interior to the chest wall. In two patients with lung
and chest wall. The dynamic scans were requested by pulmonolo- cancers adjacent to the mediastinum, the extrapulmonary mea-
38 K. Sakuma et al. / European Journal of Radiology 87 (2017) 3644

Fig. 1. Example of automatic tracking by the in-house software. A blue cross, placed in the center of the lung cancer in the right upper lobe, indicates that the tracked
center appears on this sagittal plane. A yellow cross in the adjacent rib suggests that the true tracked point in the rib is out of this sagittal plane, while the coordinates in
the y- and z-axes of the tracked point are shown at the yellow cross. Number six, which designates both tracking points, indicates that this is the sixth frame of the entire
dynamic-ventilation CT (a total of 9 frames for this patient).

suring point was placed in the thoracic aorta. In a single patient If a lung cancer does not have pleural invasion/adhesion, the respi-
with the cancer at the base of the lung, the extrapulmonary point ratory movement of the cancer is less restricted than the movement
was located below the diaphragm (the highest point of the hepatic of a tumor with pleural invasion/adhesion, resulting in larger total
dome). The software automatically tracked these two measuring distances of the cancer (Fig. 4) [8,10]. After the total distances of the
points (cancer and the adjacent structure), and coordinates were lung cancer and by the adjacent structure are calculated, the ratio
recorded in three directions (Fig. 1, Movie 1 as Supplementary of the total distances is calculated (total distance of lung cancer per
data). If the radiologist noticed that the tracked points were inac- total distance of adjacent structure).
curate because of large respiratory movements, minimum manual
corrections were also added.
2.4.3. Vector analysis
For a lung cancer that extensively invades the parietal pleura, it
2.4. Quantitative locus analysis can be predicted that not only the movement distance, but the vec-
tors of and between the tumor and the adjacent structure would
Fig. 2 demonstrates an example of 3-dimensional (3D) loci of a be similar to each other. With this assumption, the similarity of
lung cancer and an adjacent structure (rib). From these data of the each movement vector for the lung cancer and adjacent structure
3D loci, we focused on three methods of analysis that are described was expressed as cosine similarity, as follows: cosine similarity
in the following sections and quantitatively analyzed the corre- between the inspiratory and expiratory vectors (from the cancer
sponding parameters. to the adjacent structure, Fig. 5A), cosine similarity between vec-
tors of the lung cancer and adjacent structure (from the maximum
2.4.1. Changes in distance from the inspiratory frame inspiratory to the maximum expiratory frame, Fig. 5B). The cosine
After setting the maximum inspiratory frame as the standard similarity was dened by the following equation:
frame, the distance of movement by the lung cancer and the adja-
cent structure from the standard frame to each other frame was 
AB  Ai Bi
calculated, and converted to time curves of the distances moved i=1
(Fig. 3). If the lung tumor and adjacent structure move indepen- cos = =  
| A || B | 3
dently (without pleural invasion/adhesion), the two time curves  Ai  Bi
i=1 i=1
would not be correlated with each other. The similarity of the two
time curves was estimated by calculation of a cross-correlation
coefcient [4,13]. The ratio of the maximum distances from the
inspiratory frame (lung cancer per adjacent structure) was also 2.5. Statistical analysis
Continuous variables were expressed as mean standard devi-
2.4.2. Total movement distances from the rst frame ation (SD). Comparisons between the two patient groups (with
The movement distances between each of two serial frames by and without pleural invasion/adhesion) were performed using the
the lung cancer and by the adjacent structure were calculated (from Mann-Whitney U test. P values less than 0.05 were considered sta-
the rst frame), and then the distances were summed to obtain the tistically signicant. All statistical analyses were performed using
total distances moved by the cancer and by the adjacent structure. JMP 8.0 software (SAS Institute, Cary, NC).
K. Sakuma et al. / European Journal of Radiology 87 (2017) 3644 39

Fig. 2. Example of 3 dimensional (3D) loci of lung cancer and the adjacent structure (rib). In this patient (62-year-old female with adenocarcinoma in the right lower lobe), the
loci and the directions of 3D movement of the lung cancer and the rib during respiratory movements are obviously different, indicating absence of parietal pleural invasion
or adhesion. This was conrmed by a subsequent surgical procedure.

Fig. 3. Example of time curves of movement distances by the lung cancer and adjacent structure (rib) from the inspiratory frame. In this patient, the 5th frame was conrmed
as the maximum inspiratory frame and set as the standard frame. At each of the other frames, the distances moved from the standard frame by the cancer and by the rib
were calculated and plotted as time curves. In this patient, the lung cancer clearly moved a greater distance than the rib. The cross-correlation coefcient of the two time
curves was 0.852 and the ratio of the maximum distances (cancer per rib) was 4.6.

3. Results 3.2. Quantitative locus analysis

3.1. Radiation dose assessment Generally, the movements of the loci of the lung cancer and
the adjacent structure were similar in patients with parietal pleu-
For a single gantry rotation of 160 mm (0.35 s), the CT dose index ral invasion/adhesion, while they were independent in patients
volume (CTDI vol) was 1.28 mGy for the medium scanning FOV or without invasion/adhesion (Figs. 6 and 7). In addition, the lung can-
1.41 mGy for the large FOV. The dose-length product (DLP) value cer of patients without parietal pleural invasion/adhesion moved
for a single rotation was 20.4 mGy cm (medium) or 22.5 mGy cm a greater distance during respiration than the distance seen for
(large). The total estimated radiation exposure for the dynamic- cancers with parietal pleural invasion/adhesion, which showed
ventilation CT for 4.56.5 s ranged from 4.2 to 6.1 mSv (using a restricted movement. Table 1 summarizes the parameters obtained
conversion factor of 0.0145 for the chest) [14]. for the patients with and without pleural invasion/adhesion.
The analysis of the distance from the inspiratory frame
showed that the ratios of the maximum distances (tumor
40 K. Sakuma et al. / European Journal of Radiology 87 (2017) 3644

Fig. 4. Example of total movement distances of lung cancer and adjacent structure (rib). In this patient (65-year-old male, adenocarcinoma in the left lower lobe without
pleural invasion), the total distance of the cancer is approximately 8-fold greater than that of the rib, suggesting that the cancer moves freely in the thorax.

Table 1
Quantitative parameters obtained from dynamic-ventilation CT images of subpleural lung cancer patients with and without pleural invasion/adhesion.

Mean SD P value

Patients with pleural Patients without

invasion/adhesion pleural
(n = 6) invasion/adhesion
(n = 10)

Distance from inspiratory frame

Maximum distance of lung cancer (mm) 13.2 10.2 32.3 24.2 NS (0.2)
Maximum distance of adjacent structure (mm) 12.7 9.3 6.4 3.5 NS (0.3)
Ratio of the maximum distancesa 1.05 0.33 5.03 3.34 <0.01
Cross-correlation coefcientb 0.96 0.04 0.61 0.36 <0.01

Total movement distance

Total distance of lung cancer (mm) 29.2 18.3 83.6 60.2 NS (0.06)
Total distance of adjacent structure (mm) 28.5 14.4 24.8 13.7 NS (0.8)
Ratio of the total distancesc 0.97 0.27 3.30 1.58 <0.01

Vector analysis
Cosine similarity between inspiratory and expiratory vectorsd 0.80 0.27 0.08 0.63 <0.05
Cosine similarity between cancers and adjacent structures vectorse 0.96 0.05 0.61 0.33 <0.05

Abbreviation: SD = standard deviation; NS = not signicant.

The ratio of the maximum distance from the inspiratory frame of lung cancer to that of the adjacent structure.
Cross-correlation coefcient between two time curves for distances (from the inspiratory frame) of lung cancer and the adjacent structure.
The ratio of total movement distance by lung tumor to total distance by the adjacent structure.
Cosine similarity between the inspiratory and expiratory vectors, both directed from cancer to the adjacent structure.
Cosine similarity between vectors of lung cancer and the adjacent structure, both directed from inspiratory to expiratory frames.

movement/adjacent structure movement) and cross-correlation signicant between the two patient groups (P < 0.05). This result
coefcients between the two time curves (cancer and adjacent suggests that directions of the vectors were more sustained in
structure) were signicantly different for the patients with vs patients with parietal pleural invasion/adhesion than in patients
the patients without parietal pleural invasion/adhesion (P < 0.01, without invasion/adhesion.
Fig. 8). The difference in the maximum tumor diameter between
The difference in the ratios of the total movement distances the patients with and without parietal pleural invasion/adhesion
(total distance by lung cancer per total distance by adjacent struc- was also signicant (P < 0.05), suggesting that, in this study
ture) between the patients with vs the patients without parietal cohort, larger cancers were found in patients with pleural inva-
pleural invasion/adhesion was also signicant (P < 0.01). Although sion/adhesion than in the patients without invasion/adhesion.
the total movement distance by the lung cancer was greater in the Although the maximum respiratory change in the distance
patients without than in the patients with parietal pleural inva- between the cancer and the adjacent structure tended to be larger
sion/adhesion, the difference was not signicant (P = 0.06). in patients without than in patients with parietal pleural inva-
The vector analysis found that differences in both cosine simi- sion/adhesion (mean 6.5 mm with pleural invasion/adhesion and
larities (between two sets of vectors [from inspiratory to expiratory mean 21.0 mm without pleural invasion/adhesion), the difference
frames and from cancer to the adjacent structure] See Fig. 5) were was not signicant (P = 0.10).
K. Sakuma et al. / European Journal of Radiology 87 (2017) 3644 41

These novel quantitative evaluations performed using dynamic-

ventilation CT might be further developed as a new method for
evaluating parietal pleural invasion/adhesion by lung cancer.
Following the studies of visual (qualitative) assessment on
combined inspiratory and expiratory chest CT scans [8,9], cine-
MR (dynamic respiratory MR or dynamic-ventilation MR) has
been investigated for the assessment of lung cancer invasion of
the parietal pleura, chest wall, and thoracic aorta [1012,1517].
Although the accuracy and usefulness of cine-MR for identifying
tumor invasion of these adjacent organs/structures has been clearly
demonstrated in published reports, the indications of cine-MR for
lung cancer remain very limited, since preoperative chest MR for
lung cancer is generally unusual and adds to the medical costs.
Also, patients with metallic medical devices, including conven-
tional cardiac pacemakers or a cochlear implant, cannot undergo
MR imaging. After introduction of the 320-row MDCT scanner, the
potential of cine-CT of the thorax (dynamic-ventilation CT in our
terms) for diagnosing thoracic tumor invasion of adjacent struc-
tures has been reported by some investigators [57]. Although
there are no publications of direct comparisons between the accu-
racies of cine-MR and cine-CT for identifying tumor invasion of
adjacent structures, the difference in accuracy between cine-MR
and cine-CT is probably very little or absent, because both scanning
methods use similar techniques for free-breathing patients. Since
chest CT is routinely performed as a standard preoperative assess-
ment of patients with lung cancer in order to evaluate tumor size
and metastases, we currently believe that the addition of dynamic-
ventilation CT to conventional (static) chest CT for patients with
peripheral (subpleural) lung cancer is reasonable for obtaining an
accurate diagnosis of cancer invasion of the parietal pleura and
additional adjacent structures such as the aorta or pericardium.
Although both cine-MR and dynamic-ventilation CT have been
used for assessment of thoracic tumor invasion of adjacent struc-
tures, all published information has been based on the physicians
visual (qualitative) assessment [57,1012,1517]. This study was
our initial attempt at quantitative assessment of 4D data of
dynamic-ventilation CT, and an initial step in creating a new math-
ematical model for evaluation of lung cancer invasion. This model
would be incorporated into future workstations or software that
could provide completely automatic analysis for identifying tumor
Fig. 5. Illustrations for measurement of cosine similarity. Two different types of invasion of adjacent structures. We focused on three different
cosine similarity were obtained in this study. Cosine similarity between inspiratory
and expiratory vectors (shown in green arrows, from cancer to the adjacent struc-
approaches in this study, which were changes in distance moved
ture) is demonstrated in Fig. 5A. In contrast, cosine similarity between vectors of from a single frame (maximum inspiratory frame in this study),
the lung cancer and adjacent structure (from the maximum inspiratory to the max- total movement distance, and vector analysis. All three approaches
imum expiratory frame) is shown in Fig. 5B. Blue and red polygonal lines represent found that the values determined for the lung cancer and the adja-
3D movements of cancer and rib loci.
cent structure in patients with parietal pleural invasion/adhesion
were basically similar, since the target loci of the lung cancer and
4. Discussion the adjacent structure were similar in these patients. Since this is
the rst study of the quantitative analysis of 3D loci provided by
In this study, we tried to assess parietal pleural inva- dynamic-ventilation CT, there may be very different approaches
sion/adhesion by lung cancers, using several different quantitative that we did not consider. Although we think that our approaches
indices that were obtained from dynamic-ventilation CT. Using were reasonable and could comprise the standard for quantitative
the following three different approaches: (i) analysis of distance 4DCT analysis of peripheral lung cancer invasion, further quanti-
moved from the inspiratory frame, (ii) analysis of total movement tative assessments should be performed in the ongoing search for
distance, and (iii) vector analysis. We conrmed that among the better quantitative indices for the evaluation of cancer invasion.
different approaches there are several quantitative indices that Although in this study we successfully demonstrated differ-
can distinguish between patients with and without parietal pleu- ences between patients with and without parietal pleural invasion
ral invasion/adhesion. These indices include the cross-correlation in quantitative measurements obtained from dynamic-ventilation
coefcient between the time curves of distances from the inspi- CT, several methodological issues still need improvement before
ratory frame, the ratio of the total movement distances by the this technique is more widely used. For example, although we did
tumor and by the adjacent structure, and cosine similarities of vec- not perform an additional analysis, frequent observations were
tor measurements. Our ndings indicate that these quantitative made in this study that lung cancer in the upper lobe, particularly
parameters can be used to make an accurate diagnosis of tumor around the apex, did not clearly show respiratory movement, even
invasion or adhesion to the parietal pleura, and that visual (qual- in patients without parietal pleural invasion/adhesion. These nd-
itative) assessment for cancer invasion on dynamic-ventilation ings are supported by a previous study that analyzed movement
CT scans could be replaced by these quantitative assessments. of esophageal cancers on 4D-CT performed before radiotherapy
42 K. Sakuma et al. / European Journal of Radiology 87 (2017) 3644

Fig. 6. Example of a patient without parietal pleural invasion (75-year-old male, adenocarcinoma in the left lower lobe; blue line, locus of the lung cancer; pink line, locus
of the rib). The lung cancer clearly moved a greater distance than the rib, and the loci of each are different, suggesting that they move independently.

Fig. 7. Example of a patient with parietal pleural invasion (56-year-old female, adenocarcinoma in the right upper lobe; blue line, lung cancer; pink line, rib). Compared with
the patient shown in Fig. 6, the loci of the cancer and rib are to some extent similar in distance and direction of movement, suggesting that the cancer is strongly attached to
an area of the parietal pleura that is located close to the tracked rib.

and concluded that the lower esophagus moved a greater distance There are several study limitations. First, since this is a prelim-
than the upper and middle esophagus [18]. This nding might have inary study, the total number of enrolled patients was very small.
been related to respiratory movements of the diaphragm. A simi- This might account for the insignicant differences in several quan-
lar phenomenon probably occurs in lung cancer, leading to smaller titative indices. The ndings of our study must be rechecked and
respiratory movements of lung cancers located in the upper lung reproduced by additional studies with larger numbers of patients.
that would be difcult for a 4D-CT assessment of cancer invasion. Second, since this was the rst study to investigate the potential
Such problems might be minimized by training the patient to per- of dynamic-ventilation CT for assessing pleural invasion/adhesion,
form abdominal breathing for lung cancer in the lower lung and there might have been more powerful quantitative indices for dis-
costal breathing for lung cancer of the upper lung. In addition, tinguishing patients with and without pleural invasion/adhesion
the use of this 4D-CT technique for assessing cancer invasion of that we have not yet discovered. Third, we have not yet determined
adjacent structures would be extremely difcult for patients who which quantitative parameter is the best for diagnosing pleural
cannot breathe deeply because of coexisting conditions such as invasion/adhesion. Fourth, we have not created the best model for
chronic obstructive pulmonary disease, because both the lung can- including several different quantitative indices in this study. This
cer and adjacent structure move very little. This point should also should be examined in a future study enrolling a larger number of
be considered in the future when deciding on the optimal (minimal) patients. Fifth, relatively high radiation exposure was needed in the
ventilatory volume needed for performing the analysis. current study. We have started testing more reduced radiation dose
for dynamic-ventilatory scanning, such as a tube current setting of
K. Sakuma et al. / European Journal of Radiology 87 (2017) 3644 43

Fig. 8. Time curves for distances moved by the lung cancer and the adjacent structure (descending aorta) from the maximum inspiratory frame (frame 6). This patient
(79-year-old male with adenocarcinoma in the left lower lobe) had a lung cancer that had partially adhered to the descending aorta, which was conrmed surgically. In
contrast to a patient without parietal pleural invasion (See Fig. 3), the two time curves in this gure are clearly similar, suggesting that the cancer and aorta are adhered to
each other and move similarly.

20 mA. Future studies will clarify the optical tube current settings (Fumikazu Sakai, MD, PhD).
for the dynamic-ventilation CT. Sixth, assessments of the presence Shiga University of Medical Science, Otsu, Shiga, Japan;
or absence of parietal pleural invasion/adhesions were performed (Shigetaka, Sato, MD, Yukihiro Nagatani, MD, Norihisa Nitta, MD,
by different thoracic surgeons. It would have been preferable for Kiyoshi Murata, MD).
the same surgeon(s) with sufcient knowledge for the purpose of Osaka University, Suita, Osaka, Japan;
this study to perform the assessments. (Masahiro Yanagawa, MD, PhD, Osamu Honda, MD, PhD,
Noriyuki Tomiyama, MD, PhD).
5. Conclusions Osaka Medical College, Takatsuki, Osaka, Japan;
(Mitsuhiro Koyama, MD, PhD).
Dynamic-ventilation CT can be utilized as a novel imaging Tenri Hospital, Tenri, Nara, Japan;
approach for the preoperative assessment of pleural invasion and (Yuko Nishimoto, MD, Satoshi Noma, MD, PhD).
adhesion. Quantitative indices of the movements of the 3D loci of Kobe University, Kobe, Hyogo, Japan;
a lung cancer and an adjacent structure can be incorporated into a (Yoshiharu Ohno, MD, PhD).
completely automatic analysis of pleural invasion/adhesion in the Okayama University, Okayama-shi, Okayama, Japan;
future. (Katsuhide Kojima, MD).
University of the Ryukyus, Nishihara, Okinawa, Japan
Funding information (Tsuneo Yamashiro, MD, Maho Tsubakimoto, MD, Nanae
Tsuchiya MD, PhD, Sadayuki Murayama, MD, PhD).
Ohara General Hospital and University of the Ryukyus received
a research grant from Toshiba Medical Systems (Japan).
Appendix A. Supplementary data
Conict of interest
Supplementary data associated with this article can be found,
All authors do not have any personal conicts of interest to in the online version, at http://dx.doi.org/10.1016/j.ejrad.2016.12.
disclose. 004.

The authors thank Shinsuke Tsukagoshi, PhD, and Tatsuya
Kimoto, MS (Toshiba Medical Systems Corporation) for their tech- [1] S.B. Greenberg, Dynamic pulmonary CT of children, AJR Am. J. Roentgenol. 199
(2012) 435440.
nical support. The authors also thank Masayuki Hashimoto, MD [2] E.Y. Lee, E.J. Zucker, R. Restrepo, P. Daltro, P.M. Boiselle, Advanced large
(Department of Thoracic Surgery, Shiga University of Medical Sci- airway CT imaging in children: evolution from axial to 4-D assessment,
ence) for his important suggestions. Pediatr. Radiol. 43 (2013) 285297.
[3] J.Z. Tan, M. Crossett, M. Ditcheld, Dynamic volumetric computed
The ACTIve Study Group currently consists of the following insti- tomographic assessment of the young paediatric airway: initial experience of
tutions: rapid non-invasive, four-dimensional technique, J. Med. Imaging Radiat.
Ohara General Hospital, Fukushima-City, Fukushima, Japan; Oncol. 57 (2013) 141148.
[4] T. Yamashiro, H. Moriya, M. Tsubakimoto, S. Matsuoka, S. Murayama,
(Kotaro Sakuma, MD, Hiroshi Moriya, MD, PhD). Continuous quantitative measurement of the proximal airway dimensions
Saitama International Medical Center, Saitama Medical Univer- and lung density on four-dimensional dynamic-ventilation CT in smokers, Int.
sity, Hidaka, Saitama, Japan; J. Chron. Obstruct. Pulmon. Dis. 11 (2016) 755764.
44 K. Sakuma et al. / European Journal of Radiology 87 (2017) 3644

[5] C.K. Choong, S.S. Pasricha, X. Li, P. Briggs, S. Ramdave, M. Crossett, J.M. [12] S. Akata, N. Kajiwara, J. Park, M. Yoshimura, D. Kakizaki, K. Abe, T. Hirano, T.
Troupis, Dynamic four-dimensional computed tomography for preoperative Ohira, M. Tsuboi, H. Kato, Evaluation of chest wall invasion by lung cancer
assessment of lung cancer invasion into adjacent structures, Eur. J. using respiratory dynamic MRI, J. Med. Imaging Radiat. Oncol. 52 (2008)
Cardiothorac. Surg. 47 (2015) 239243. 3639.
[6] R. Fujimoto, M. Sato, R. Miyata, K. Minakata, M. Omasa, T. Kubo, H. Date, [13] T. Yamashiro, M. Tsubakimoto, Y. Nagatani, H. Moriya, K. Sakuma, S.
Successful resection of recurrent mediastinal liposarcoma using preoperative Tsukagoshi, H. Inokawa, T. Kimoto, R. Teramoto, S. Murayama, Automated
evaluation of organ invasion by four-dimensional computed tomography, continuous quantitative measurement of proximal airways on dynamic
Gen. Thorac. Cardiovasc. Surg. (December (21)) (2014) (Epub ahead of print). ventilation CT: initial experience using an ex-vivo porcine lung phantom, Int.
[7] J.M. Troupis, S.S. Pasricha, H. Narayanan, F.J. Rybicki, A.W. Pick, 4D CT and lung J. Chron. Obstruct. Pulmon. Dis. 10 (2015) 20452054.
cancer surgical resectability: a technical innovation, J. Med. Imaging Radiat. [14] P.D. Deak, Y. Smal, W.A. Kalender, Multisection CT protocols: sex- and
Oncol. 58 (2014) 469471. age-specic conversion factors used to determine effective dose from
[8] K. Murata, M. Takahashi, M. Mori, K. Shimoyama, A. Mishina, S. Fujino, H. Itoh, dose-length product, Radiology 257 (2010) 158166.
R. Morita, Chest wall and mediastinal invasion by lung cancer: evaluation [15] N. Kajiwara, S. Akata, O. Uchida, J. Usuda, T. Ohira, N. Kawate, N. Ikeda, Cine
with multisection expiratory dynamic CT, Radiology 191 (1994) MRI enables better therapeutic planning than CT in cases of possible lung
251255. cancer chest wall invasion, Lung Cancer 69 (2010) 203208.
[9] T. Shirakawa, K. Fukuda, Y. Miyamoto, H. Tanabe, S. Tada, Parietal pleural [16] Y.J. Hong, J. Hur, H.J. Lee, Y.J. Kim, S.R. Hong, Y.J. Suh, B.W. Choi, Respiratory
invasion of lung masses: evaluation with CT performed during deep dynamic magnetic resonance imaging for determining aortic invasion of
inspiration and expiration, Radiology 192 (1994) 809811. thoracic neoplasms, J. Thorac. Cardiovasc. Surg. 148 (2014) 644650.
[10] S. Sakai, S. Murayama, J. Murakami, N. Hashiguchi, K. Masuda, Bronchogenic [17] H. Uramoto, H. Kinoshita, Y. Nakajima, H. Akiyama, Easy diagnosis of aortic
carcinoma invasion of the chest wall: evaluation with dynamic cine MRI invasion in patients with lung cancer using cine magnetic resonance imaging,
during breathing, J. Comput. Assist. Tomogr. 21 (1997) 595600. Case Rep. Oncol. 8 (2015) 308311.
[11] S. Shiotani, K. Sugimura, M. Sugihara, H. Kawamitsu, M. Yamauchi, M. [18] H. Yamashita, S. Kida, A. Sakumi, A. Haga, S. Ito, T. Onoe, K. Okuma, K. Ino, M.
Yoshida, T. Kushima, T. Kinoshita, H. Endoh, H. Nakayama, M. Kajitani, M. Akahane, K. Ohtomo, K. Nakagawa, Four-dimensional measurement of the
Wada, Diagnosis of chest wall invasion by lung cancer: useful criteria for displacement of internal ducial markers during 320-multislice computed
exclusion of the possibility of chest wall invasion with MR imaging, Radiat. tomography scanning of thoracic esophageal cancer, Int. J. Radiat. Oncol. Biol.
Med. 18 (2000) 283290. Phys. 79 (2011) 588595.