respiratory investigation 52 (2014) 28 –35
Contents lists available at ScienceDirect
Respiratory Investigation journal homepage: www.elsevier.com/locate/resinv
Original Article
Vibration response imaging in healthy Japanesesubjects Masamichi Mineshitan, Taeko Shirakawa, Junko Saji, Hiroshi Handa, Naoki Furuya, Hirotaka Kida, Hiroki Nishine, Seiichi Nobuyama, Takeo Inoue, Teruomi Miyazawa Division of Respiratory and Infectious Diseases, Department of Internal Medicine, St. Marianna University School of Medicine, Kawasaki, Japan
ar t ic l e in f o
abs tra ct
Article history:
Background: Vibration response imaging (VRI) records the intensity and distribution of lung
Received 29 July 2012
sounds during the respiration cycle. Our objective was to analyze VRI findings in healthy
Received in revised form
Japanese adults.
26 May 2013
Methods: VRI images of 106 healthy subjects (33.779.6 years, 52 male and 54 female),
Accepted 30 May 2013
including 67 nonsmokers and 39 asymptomatic smokers, were recorded. The regional
Available online 6 July 2013
intensity of vibrations was assessed using quantitative lung data (QLD), and VRI dynamic
Keywords:
images by rater assessment, left and right lung asynchrony (gap index), and regional lung
Breath sounds
asynchrony (asynchrony score).
Healthy subjects
Results: A dominance of total left lung QLD was observed in all subjects, and this phenom-
Lung function
enon was more prominent in female subjects. However, there was no significant difference
Vibration response imaging
between the total left and total right lung QLD in smokers. Rater assessments showed that 81.1% of all subjects had a normal final assessment. Male subjects had a significantly higher percentage of good or normal assessments for all image scores, except dynamic image scoring. The asynchrony score was significantly higher in female subjects. There were no significant differences in these qualitative assessments between non-smokers and smokers. Conclusions: Although our QLD results were similar to those of a previous report, there were discrepancies between sexes for the qualitative assessments. A significantly higher number of female subjects had abnormal images as assessed by the raters. Furthermore, significantly higher asynchrony scores were observed in female subjects. The VRI variability in sex may be considered normal among the Japanese population. This study is registered with UMIN-CTR under registration number UMIN000002355. & 2013 The Japanese Respiratory Society. Published by Elsevier B.V. All rights reserved.
n
Correspondence to: Division of Respiratory and Infectious Diseases, Department of Internal Medicine, St. Marianna University School of Medicine, 2-16-1 Sugao, Miyamae-ku, Kawasaki-City, Kanagawa 216-8511, Japan. Tel.: +81 44 977 8111; fax: +81 44 977 8361. E-mail addresses: [email protected].(M. Mineshita)[email protected].(T. Shirakawa)[email protected] .(J. Saji)[email protected].(H. Handa)[email protected].(N. Furuya)[email protected].(H. Kida) [email protected].(H. Nishine)[email protected].(S. Nobuyama)[email protected].(T. Inoue) [email protected].(T. Miyazawa) 2212-5345/$ - see front matter & 2013 The Japanese Respiratory Society. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.resinv.2013.05.008
respiratory investigation 52 (2014) 28 –35
1.
Introduction
Since the invention of the stethoscope by Laennec in 1816, auscultation has frequently provided important diagnostic information. Physicians have routinely listened to the sounds produced by a patient's internal organs, such as the heart and lungs, as an aid in diagnosing and treating various disorders. Lung sounds have also been used to assess ventilation. It is generally accepted that louder breath sounds indicate higher ventilation in healthy subjects. Ploy-Song-Sang [1] reported that regional breath sounds are a fair index of regional ventilation in subjects in an upright posture. However, the stethoscope has limitations. Auscultation is a subjective process that is dependent on the auditory acuity and clinical experience of the user. Furthermore, it does not provide a quantitative evaluation of breath sounds. Vibration response imaging (VRI) is a commercially developed acoustic lung imaging system that displays breath sound distributions as dynamic gray-scale images [2,3]. Vibrational energy is acquired during a 12-s recording, and the algorithm extracts the relative intensities of the vibrations for each lung and each lung region (upper, middle, and lower). The relative intensities of the vibrations in the 12-s recording are then aggregated and expressed as a percentage of the total for both lungs combined (quantitative lung data [QLD]) [4]. A recent study established that the QLD obtained using VRI were highly reproducible for recordings performed on the same individual at different time intervals [5]. Using the VRI method, Yigla [4] reported characteristic features of VRI and QLD from healthy asymptomatic subjects and speculated that VRI dynamic images and QLD might be able to detect subtle disease changes in cigarette smokers who were observed to be otherwise healthy. In other reports, VRI has been applied to diagnose obstructive airway diseases [6–9]. To supplement VRI dynamic image data, Wang [6] introduced gaps in vibration energy peaks (VEPs) to evaluate the asynchrony between the airflow through the left and right lung in asthma exacerbations, which was significantly reduced following treatment. The introduction of mathematical analyses for VRI appears to provide objective physiological information and may be helpful for users unfamiliar with VRI assessments. Sound transmission from the airways to the chest surface is influenced by various factors, including lung structure and chest wall configuration. For example, body size and chest wall dimensions have been reported to affect respiratory sounds [10,11]. Because Japanese subjects tend to have smaller body sizes and thinner thoraxes than European subjects, the VRI findings using Japanese subjects may be different from Yigla's data, which were based on Israeli subjects [4]. In this study, we recorded VRI measurements of healthy Japanese subjects. The data were analyzed by trained raters, and mathematical measurements were obtained. The results were compared with previous findings [4].
2.
Materials and methods
2.1.
Subjects
Fifty-seven male (22–57 years) and 56 female (21–60 years) healthy volunteers were recruited for this study. Volunteers were
29
deemed healthy on the basis of their clinical history, a physical examination, and spirometric findings. Because a chest X-ray was obtained during annual health checks for all subjects, they were not repeated, thus avoiding additional radiation exposure. Subjects whose history included abnormal chest X-ray findings in the previous year were excluded. Individuals with a history of chronic cardiopulmonary disease, surgical chest procedures, or a recent (within the preceding 6 months) respiratory tract infection were also excluded. Spirometry was performed using a calibrated spirometer (HI-801; Chest M.I., Inc., Tokyo, Japan). Prior to testing, calibration checks were performed using a 3 L calibration syringe with ambient air. The predicted values of the spirometric measurements were derived from the Japanese Respiratory Society [12]. The ethics committee of the St. Marianna University School of Medicine approved this study, and written consent was obtained from all participants (Approved date: May 1, 2007; Approved #: 1230). This study is registered with UMIN-CTR under registration number UMIN000002355.
2.2.
VRI System
The VRIxp System (Deep Breeze, Or-Akiva, Israel) is a computerbased acoustic lung imaging platform developed to acquire, quantify, monitor, and store breath sounds [2,3]. The equipment was provided by the Deep Breeze Co. Recordings were performed using the VRI device, as previously described [2–4]. Generally, the recordings were performed with 7-row arrays (40 active sensors in total); however, we used 6-row arrays (34 active sensors in total) when the subject's height was below 165 cm. The first rows of sensors on the left and right peripheries were inactive. The signal data were band-pass filtered (100–250 Hz) to reduce the interference generated by chest wall movements and heart sounds. The algorithm expresses the data quantitatively and maps the locations of the detected sound vibrations. The filtered frequencies are represented graphically by a gray scale-coded dynamic image based on a 2-dimensional coordinate system that corresponds to the regional position of the sensors (Supplementary Content). High-intensity vibrational energy is depicted as dark gray to black. Low-intensity vibrational energy is graycoded, and the absence of vibration appears white. In the dynamic image, the distribution of the vibrational energy is displayed along a timeline for inspiration and expiration. Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.resinv.2013.05.008. The sum of the signal energy for each row of the left and right planar arrays over the full 12 s recording was used to compute an overall average QLD in each lung field. The data were further divided into the upper (rows 1–2), middle (rows 3–4 in the 6-row arrays and rows 3–5 in the 7-row arrays) and lower (rows 5–6 in the 6-row arrays and rows 6–7 in the 7-row arrays) zones, and QLD values for each of these 6 zones were generated. The QLD variable is continuous and ranges between 0% and 100%. The QLD data are reported as the mean7standard deviation (SD).
2.3.
Recording procedure
Subjects were seated in a quiet environment with their hands resting in their laps. Left and right planar arrays were placed on the subject's back. The arrays were positioned symmetrically with the upper row of each array approximately 2 cm above the
30
respiratory investigation 52 (2014) 28 –35
scapula and the inner sensors of the upper rows approximately 5 cm from the vertebral column. The two bottom row arrays were approximately at the same height as the two arrays parallel to the vertebral column. Recordings were performed over a 12-s period while the subjects took deep regular breaths at a rate of 15–20 breaths/min. The investigator instructed the subjects to target their breathing to a range of 1.5–3.5 on the breathing intensity bar on the VRI monitor and to a breathing rate of 15– 20 cycles/min. Each subject was recorded at least 3 times, and 1 acceptable VRI recording with the highest technical quality was chosen using the following criteria: 1. Recordings with artifacts were excluded. 2. The breathing cycles (3–4 cycles per 12 s) were within the requested range. 3. The breathing intensity was adequate and showed the most consistent breathing pattern.
2.4.
Assessment of VRI dynamic images
The VRI dynamic images were assessed using the following methods: (1) qualitative assessment; (2) left and right lung asynchrony; and (3) regional lung asynchrony.
2.4.1.
Qualitative assessment of VRI images
Three trained raters analyzed the dynamic images. Raters were blinded to the subjects and the scores given by the other raters. The raters evaluated each image for the following features [4]:
Dynamic image (good/disturbed). Image development (good/medium/poor). Maximum-Energy-Frame (MEF) shape (good/medium/ poor).
MEF area (R4L; R¼ L; RoL). MEF intensity (R4L; R¼ L; RoL). MEF missing parts (yes/no). Four locations were evaluated
for MEF missing parts: the right upper, left upper, right lower, and left lower quadrants. Each location was rated depending on the presence or absence of a part normally expected to appear. Final assessment (normal/abnormal)
A normal dynamic image moves vertically and centrifugally first and then moves in the opposite direction vertically and centripetally. This pattern is synchronized in the left and right lungs. The image becomes larger from one frame to the next (progression) and then becomes smaller from one frame to the next (regression). Good image development includes consistent progression and regression. The MEF is the frame in which the gray-scale distribution of the lungs is at a maximum and the data values are the highest. The final assessment was determined as described in Yigla's study [4]. For qualitative assessment, the raters' evaluations were coded and analyzed to characterize the healthy population. The characteristic features were validated based on the inter-
rater agreement (the degree to which the raters were consistent with each other for the same image). For each subject and feature, the number of evaluations that appeared most often (i.e., the mode) across the raters was calculated (i.e., the frequency of the mode¼f (mode)). For each subject and for all features, the sum of the f (mode) values was calculated (Σf (mode)). Normalization was performed to generate agreement levels between 0% and 100% (0%¼ no agreement; 100%¼ full agreement). Normalization was obtained by subtracting the minimum Σf (mode) that could be achieved (16) and dividing the result by the difference between the maximum and minimum Σf (mode) that could be achieved (30−16¼ 14). The average inter-rater agreement for all subjects was computed using 100 n 10 n ∑10 i ¼ 1 f ðmodeÞ−Minð∑i ¼ 1 f ðmodeÞÞ ∑ 10 10 J ¼ 1 Maxð∑i ¼ 1 f ðmodeÞÞ−Minð∑i ¼ 1 f ðmodeÞÞ
Inter rater agreement ð%Þ ¼
i ¼ 1; ::; 10 features j ¼ 1; ::; n subjects
ð1Þ
The suggested benchmarks for reliability are as follows: poor to slight, 0–20%; fair, 21–40%; moderate, 41–60%; substantial, 61–80%; and excellent, 81–100% [13].
2.4.2.
Left and right lung asynchrony
The vibrational energy in dB can be plotted over time for each lung, with time on the x-axis and dB on the y-axis (Fig. 1). The timing of the inspiratory and expiratory VEPs was compared for the left and right lungs [6]. The gap index was defined as the absolute value of the average of the gaps (s) between the VEPs of both lungs for 12 s.
2.4.3.
Regional lung asynchrony
Although whole left lung–right lung asynchrony was confirmed by the gap index, regional left lung–right lung asynchrony might have been masked if there were differences in the time course of regional airflow. The time course of vibrational energy for each row (6–7 rows for each lung), in addition to that for the whole lung, can be plotted and used to analyze regional lung asynchrony (Fig. 2). Because VRI collects data 71 times in 12 s, the recording period can be divided into 70 sections of 0.17 s. We defined the asynchronous section in each row as the section in which the change (increase or decrease) in vibrational energy of the right and left rows was discordant. The asynchrony score is defined as the average number of asynchronous sections in each row.
2.5.
Statistical analysis
A p-value of o0.05 was considered significant for all tests. Significant differences in the regional lung QLD between the left and right lungs were evaluated using two-sided t-tests for paired data. Differences in the regional QLD, the gap index, and the asynchrony score between male and female subjects were also evaluated using two-sided t-tests for unpaired data. The qualitative assessments from the raters were analyzed with comparisons between the male and female subjects and between the non-smokers and smokers by using chi-squared tests.
31
respiratory investigation 52 (2014) 28 –35
Gap between vibration energy peaks (VEP) of lungs Left
Time course of vibration energy(frame)
Right
Vibration energy (dB)
-35
Left lung
Right lung
-40
VEP gap -45 -50 -55 -60 -65
inspiration
expiration
the gap index = |average of VEP gap (inspiratory and expiratory) | Fig. 1 – Gaps between the vibrational energy peaks (VEP) of the lungs. The vibrational energy in dB can be plotted over time for each lung, with time on the x-axis and dB on the y-axis. The timing of the inspiratory and expiratory VEPs was compared for the left and right lungs. The gap index was defined as the absolute value of the average of the gaps (s) between the VEPs of both lungs for a duration of 12 s.
Regional left and right lung asynchrony Time course of vibration energy In bilateral 1st row (frame) Right 1st row
•• •• ••
• • •
-40
Vibration energy (dB)
Left 1st row
-45 -50 -55 -60 -65 -70 -75 Lt 1st row
Rt 1st row
●●
●
the asynchrony score (AS) = average number of the asynchronous sections of each row
●: asynchronous section Fig. 2 – Regional left and right lung asynchrony. The time course of vibrational energy for each row can be plotted and used to analyze regional lung asynchrony. Because VRI collects data 71 times during 12 s, the recording period can be divided into 70 sections of 0.17 s. We defined the asynchronous section in each row as the section at which the change (increase or decrease) in vibrational energy of the left and right rows was discordant. The asynchrony score is defined as the average number of asynchronous sections for each row.
3.
Results
3.1.
Characteristics of the subjects
Recordings were made of 57 male and 56 female subjects. One female and 2 male subjects were excluded during the screening stage for not meeting the eligibility criteria (1 with asthma and 2 diagnosed with restrictive lung disease after the lung
function tests). One male subject was excluded because of missing lung function test results, 2 male subjects were excluded because of missing inter-rater agreement data, and 1 female subject was excluded because of image artifacts. The remaining 106 subjects (52 male and 54 female; mean age, 33.779.6 years), including 67 non-smokers and 39 smokers (11 ex-smokers and 28 current smokers), were included in this study.
32
respiratory investigation 52 (2014) 28 –35
The demographics, anthropometric values, and lung function test results of the study population are shown in Table 1. The smoking rates in female subjects were significantly lower than those in male subjects (po0.005). Male smokers were significantly older than male non-smokers (po0.05). Male smokers had significantly lower forced expiratory volume in 1 s (FEV1)/forced vital capacity (FVC) compared with male non-smokers (po0.05).
3.2.
QLD analysis
The average total left QLD value, including the upper, middle, and lower QLD values, was significantly higher than the corresponding value in the right lung for all subjects, with the exception of the lower regions in male subjects (Table 2). The dominance of left lung QLD was more prominent in female subjects. The average right upper, middle, and total QLD values were significantly lower, whereas the average left lower and total QLD values were significantly higher in female subjects. Subgroup analysis in non-smokers showed that the dominance of the total left lung QLD was significant in all subjects (Table 2). However, there were no significant differences between the total left and right lung QLD values in smokers (Table 2).
3.3.
Qualitative assessment of VRI images
The overall inter-rater agreement for all features was 82.9%. For the final assessment, 81.1% of all subjects were considered normal. The ratios for good dynamic image, good image development, good MEF shape, symmetric MEF area, symmetric MEF intensity, and no missing parts were 87.7%, 74.5%, 85.8%, 84.9%, 78.3%, and 91.5%, respectively. The final assessments and all parameter scores for the images, except for the dynamic images, were significantly lower in female subjects (Fig. 3). We analyzed inspiratory peaks in 106 subjects and expiratory peaks in 102 subjects because expiratory peaks were not distinct for 2 male non-smokers and 2 female non-smokers. The average gap index of the 106 subjects was 0.0670.05 s,
and the gap index was lower than 0.06 s in 76 subjects (71.7%). The average asynchrony score for all subjects was 10.873.5. Although no significant difference between the sexes was observed in the gap index (female subjects, 0.0670.06 s; male subjects, 0.0570.04 s), there was a significant difference in the asynchrony score between female and male subjects (female subjects, 12.573.6; male subjects, 9.172.4; po0.001, Fig. 4). We did not find any significant differences in the qualitative assessments between the non-smokers and the smokers (Table 3).
4.
Discussion
Our QLD results were similar to those of a previously published report [4]. However, there were discrepancies in the effects of sex on the qualitative assessment. A significantly greater number of female subjects in our study had abnormal images, as judged by the raters. The mathematical analyses we introduced for VRI assessment also revealed significantly higher regional lung asynchrony in female subjects. Because this phenomenon was not found in a previous study of Israeli subjects [4], the VRI variability in sex demonstrated in this study may be considered as a normal characteristic among the Japanese population. VRI is a commercially developed acoustic lung imaging system that displays breath sound distributions as dynamic gray-scale images [2,3]. Reports in which VRI was applied in clinical settings, such as in interventional bronchoscopy, lung transplantation, pleural effusion, pneumothorax, foreign body aspiration, and obstructive lung diseases, have been published [6–9,14–18]. Yigla [4] reported the characteristic features of VRI in healthy asymptomatic subjects; the researchers reported that 89% of the healthy population had a normal final assessment. The total left lung QLD value was significantly higher than the right lung QLD value. This left lung dominance in QLD was more prominent in female subjects and was weaker in smokers. Although the subgroup analysis did not show any significant effects of sex using the qualitative assessment, a significantly higher proportion of heavy smokers had images rated as abnormal.
Table 1 – Characteristics of subjects. Male
No. of subjects Brinkman index Pack-years Age (y) Height (cm) Weight (kg) BMI FVC (% predicted) FEV1 (% predicted) FEV1/FVC (%)
Female
All
Non-smokers
Smokers
All
Non-smokers
Smokers
52 151.57217.8 7.6711.0 34.379.1 171.975.2 69.679.5 23.572.8 99.878.6 95.378.2 83.975.3
25 0 0 30.977.4 171.875.0 68.6710.3 23.273.0 97.278.1 94.677.2 86.275.4
27 303.07221.5 15.2711.1 36.178.7* 171.975.4 70.778.8 23.972.7 102.178.8 96.079.2 82.374.3**
54 30.4776.4 1.573.8 33.8710.7 158.375.1 53.377.0 21.372.8 102.9711.8 103.8711.3 87.475.9
42 0 0 34.2711.7 157.875.0 52.476.9 21.072.4 102.8712.8 103.9711.1 87.775.8
12 136.87110.9 6.875.5 32.175.8 159.875.7 56.376.5 22.274.0 103.478.0 103.4712.5 86.576.3
Definition of abbreviation: BMI ¼body mass index, FVC ¼forced vital capacity and FEV1 ¼ forced expiratory in 1 s. Values are represented as means7standard deviation. n po0.05. nn po0.01. Compared with non-smoker.
33
6.773.4 14.575.3 24.575.8 45.779.0 8.972.1# 20.876.1# 24.677.2 54.379.0
Smokers n ¼12
respiratory investigation 52 (2014) 28 –35
(%) 100
Qualitative assessment of VRI images *
80 60
*
*
* *
*
5.973.2 13.373.7 21.475.5 40.677.0 10.272.6### 22.874.6### 26.574.8### 59.477.0###
20
6.173.2ǁǁ 13.674.1ǁǁǁ 22.175.7 41.777.7ǁǁǁ 9.972.5### 22.375.0### 26.175.4#ǁǁǁ 58.377.7###ǁǁǁ
0
Male (n=52) Female (n=54)
*p
Contents lists available at ScienceDirect
Respiratory Investigation journal homepage: www.elsevier.com/locate/resinv
Original Article
Vibration response imaging in healthy Japanesesubjects Masamichi Mineshitan, Taeko Shirakawa, Junko Saji, Hiroshi Handa, Naoki Furuya, Hirotaka Kida, Hiroki Nishine, Seiichi Nobuyama, Takeo Inoue, Teruomi Miyazawa Division of Respiratory and Infectious Diseases, Department of Internal Medicine, St. Marianna University School of Medicine, Kawasaki, Japan
ar t ic l e in f o
abs tra ct
Article history:
Background: Vibration response imaging (VRI) records the intensity and distribution of lung
Received 29 July 2012
sounds during the respiration cycle. Our objective was to analyze VRI findings in healthy
Received in revised form
Japanese adults.
26 May 2013
Methods: VRI images of 106 healthy subjects (33.779.6 years, 52 male and 54 female),
Accepted 30 May 2013
including 67 nonsmokers and 39 asymptomatic smokers, were recorded. The regional
Available online 6 July 2013
intensity of vibrations was assessed using quantitative lung data (QLD), and VRI dynamic
Keywords:
images by rater assessment, left and right lung asynchrony (gap index), and regional lung
Breath sounds
asynchrony (asynchrony score).
Healthy subjects
Results: A dominance of total left lung QLD was observed in all subjects, and this phenom-
Lung function
enon was more prominent in female subjects. However, there was no significant difference
Vibration response imaging
between the total left and total right lung QLD in smokers. Rater assessments showed that 81.1% of all subjects had a normal final assessment. Male subjects had a significantly higher percentage of good or normal assessments for all image scores, except dynamic image scoring. The asynchrony score was significantly higher in female subjects. There were no significant differences in these qualitative assessments between non-smokers and smokers. Conclusions: Although our QLD results were similar to those of a previous report, there were discrepancies between sexes for the qualitative assessments. A significantly higher number of female subjects had abnormal images as assessed by the raters. Furthermore, significantly higher asynchrony scores were observed in female subjects. The VRI variability in sex may be considered normal among the Japanese population. This study is registered with UMIN-CTR under registration number UMIN000002355. & 2013 The Japanese Respiratory Society. Published by Elsevier B.V. All rights reserved.
n
Correspondence to: Division of Respiratory and Infectious Diseases, Department of Internal Medicine, St. Marianna University School of Medicine, 2-16-1 Sugao, Miyamae-ku, Kawasaki-City, Kanagawa 216-8511, Japan. Tel.: +81 44 977 8111; fax: +81 44 977 8361. E-mail addresses: [email protected].(M. Mineshita)[email protected].(T. Shirakawa)[email protected] .(J. Saji)[email protected].(H. Handa)[email protected].(N. Furuya)[email protected].(H. Kida) [email protected].(H. Nishine)[email protected].(S. Nobuyama)[email protected].(T. Inoue) [email protected].(T. Miyazawa) 2212-5345/$ - see front matter & 2013 The Japanese Respiratory Society. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.resinv.2013.05.008
respiratory investigation 52 (2014) 28 –35
1.
Introduction
Since the invention of the stethoscope by Laennec in 1816, auscultation has frequently provided important diagnostic information. Physicians have routinely listened to the sounds produced by a patient's internal organs, such as the heart and lungs, as an aid in diagnosing and treating various disorders. Lung sounds have also been used to assess ventilation. It is generally accepted that louder breath sounds indicate higher ventilation in healthy subjects. Ploy-Song-Sang [1] reported that regional breath sounds are a fair index of regional ventilation in subjects in an upright posture. However, the stethoscope has limitations. Auscultation is a subjective process that is dependent on the auditory acuity and clinical experience of the user. Furthermore, it does not provide a quantitative evaluation of breath sounds. Vibration response imaging (VRI) is a commercially developed acoustic lung imaging system that displays breath sound distributions as dynamic gray-scale images [2,3]. Vibrational energy is acquired during a 12-s recording, and the algorithm extracts the relative intensities of the vibrations for each lung and each lung region (upper, middle, and lower). The relative intensities of the vibrations in the 12-s recording are then aggregated and expressed as a percentage of the total for both lungs combined (quantitative lung data [QLD]) [4]. A recent study established that the QLD obtained using VRI were highly reproducible for recordings performed on the same individual at different time intervals [5]. Using the VRI method, Yigla [4] reported characteristic features of VRI and QLD from healthy asymptomatic subjects and speculated that VRI dynamic images and QLD might be able to detect subtle disease changes in cigarette smokers who were observed to be otherwise healthy. In other reports, VRI has been applied to diagnose obstructive airway diseases [6–9]. To supplement VRI dynamic image data, Wang [6] introduced gaps in vibration energy peaks (VEPs) to evaluate the asynchrony between the airflow through the left and right lung in asthma exacerbations, which was significantly reduced following treatment. The introduction of mathematical analyses for VRI appears to provide objective physiological information and may be helpful for users unfamiliar with VRI assessments. Sound transmission from the airways to the chest surface is influenced by various factors, including lung structure and chest wall configuration. For example, body size and chest wall dimensions have been reported to affect respiratory sounds [10,11]. Because Japanese subjects tend to have smaller body sizes and thinner thoraxes than European subjects, the VRI findings using Japanese subjects may be different from Yigla's data, which were based on Israeli subjects [4]. In this study, we recorded VRI measurements of healthy Japanese subjects. The data were analyzed by trained raters, and mathematical measurements were obtained. The results were compared with previous findings [4].
2.
Materials and methods
2.1.
Subjects
Fifty-seven male (22–57 years) and 56 female (21–60 years) healthy volunteers were recruited for this study. Volunteers were
29
deemed healthy on the basis of their clinical history, a physical examination, and spirometric findings. Because a chest X-ray was obtained during annual health checks for all subjects, they were not repeated, thus avoiding additional radiation exposure. Subjects whose history included abnormal chest X-ray findings in the previous year were excluded. Individuals with a history of chronic cardiopulmonary disease, surgical chest procedures, or a recent (within the preceding 6 months) respiratory tract infection were also excluded. Spirometry was performed using a calibrated spirometer (HI-801; Chest M.I., Inc., Tokyo, Japan). Prior to testing, calibration checks were performed using a 3 L calibration syringe with ambient air. The predicted values of the spirometric measurements were derived from the Japanese Respiratory Society [12]. The ethics committee of the St. Marianna University School of Medicine approved this study, and written consent was obtained from all participants (Approved date: May 1, 2007; Approved #: 1230). This study is registered with UMIN-CTR under registration number UMIN000002355.
2.2.
VRI System
The VRIxp System (Deep Breeze, Or-Akiva, Israel) is a computerbased acoustic lung imaging platform developed to acquire, quantify, monitor, and store breath sounds [2,3]. The equipment was provided by the Deep Breeze Co. Recordings were performed using the VRI device, as previously described [2–4]. Generally, the recordings were performed with 7-row arrays (40 active sensors in total); however, we used 6-row arrays (34 active sensors in total) when the subject's height was below 165 cm. The first rows of sensors on the left and right peripheries were inactive. The signal data were band-pass filtered (100–250 Hz) to reduce the interference generated by chest wall movements and heart sounds. The algorithm expresses the data quantitatively and maps the locations of the detected sound vibrations. The filtered frequencies are represented graphically by a gray scale-coded dynamic image based on a 2-dimensional coordinate system that corresponds to the regional position of the sensors (Supplementary Content). High-intensity vibrational energy is depicted as dark gray to black. Low-intensity vibrational energy is graycoded, and the absence of vibration appears white. In the dynamic image, the distribution of the vibrational energy is displayed along a timeline for inspiration and expiration. Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.resinv.2013.05.008. The sum of the signal energy for each row of the left and right planar arrays over the full 12 s recording was used to compute an overall average QLD in each lung field. The data were further divided into the upper (rows 1–2), middle (rows 3–4 in the 6-row arrays and rows 3–5 in the 7-row arrays) and lower (rows 5–6 in the 6-row arrays and rows 6–7 in the 7-row arrays) zones, and QLD values for each of these 6 zones were generated. The QLD variable is continuous and ranges between 0% and 100%. The QLD data are reported as the mean7standard deviation (SD).
2.3.
Recording procedure
Subjects were seated in a quiet environment with their hands resting in their laps. Left and right planar arrays were placed on the subject's back. The arrays were positioned symmetrically with the upper row of each array approximately 2 cm above the
30
respiratory investigation 52 (2014) 28 –35
scapula and the inner sensors of the upper rows approximately 5 cm from the vertebral column. The two bottom row arrays were approximately at the same height as the two arrays parallel to the vertebral column. Recordings were performed over a 12-s period while the subjects took deep regular breaths at a rate of 15–20 breaths/min. The investigator instructed the subjects to target their breathing to a range of 1.5–3.5 on the breathing intensity bar on the VRI monitor and to a breathing rate of 15– 20 cycles/min. Each subject was recorded at least 3 times, and 1 acceptable VRI recording with the highest technical quality was chosen using the following criteria: 1. Recordings with artifacts were excluded. 2. The breathing cycles (3–4 cycles per 12 s) were within the requested range. 3. The breathing intensity was adequate and showed the most consistent breathing pattern.
2.4.
Assessment of VRI dynamic images
The VRI dynamic images were assessed using the following methods: (1) qualitative assessment; (2) left and right lung asynchrony; and (3) regional lung asynchrony.
2.4.1.
Qualitative assessment of VRI images
Three trained raters analyzed the dynamic images. Raters were blinded to the subjects and the scores given by the other raters. The raters evaluated each image for the following features [4]:
Dynamic image (good/disturbed). Image development (good/medium/poor). Maximum-Energy-Frame (MEF) shape (good/medium/ poor).
MEF area (R4L; R¼ L; RoL). MEF intensity (R4L; R¼ L; RoL). MEF missing parts (yes/no). Four locations were evaluated
for MEF missing parts: the right upper, left upper, right lower, and left lower quadrants. Each location was rated depending on the presence or absence of a part normally expected to appear. Final assessment (normal/abnormal)
A normal dynamic image moves vertically and centrifugally first and then moves in the opposite direction vertically and centripetally. This pattern is synchronized in the left and right lungs. The image becomes larger from one frame to the next (progression) and then becomes smaller from one frame to the next (regression). Good image development includes consistent progression and regression. The MEF is the frame in which the gray-scale distribution of the lungs is at a maximum and the data values are the highest. The final assessment was determined as described in Yigla's study [4]. For qualitative assessment, the raters' evaluations were coded and analyzed to characterize the healthy population. The characteristic features were validated based on the inter-
rater agreement (the degree to which the raters were consistent with each other for the same image). For each subject and feature, the number of evaluations that appeared most often (i.e., the mode) across the raters was calculated (i.e., the frequency of the mode¼f (mode)). For each subject and for all features, the sum of the f (mode) values was calculated (Σf (mode)). Normalization was performed to generate agreement levels between 0% and 100% (0%¼ no agreement; 100%¼ full agreement). Normalization was obtained by subtracting the minimum Σf (mode) that could be achieved (16) and dividing the result by the difference between the maximum and minimum Σf (mode) that could be achieved (30−16¼ 14). The average inter-rater agreement for all subjects was computed using 100 n 10 n ∑10 i ¼ 1 f ðmodeÞ−Minð∑i ¼ 1 f ðmodeÞÞ ∑ 10 10 J ¼ 1 Maxð∑i ¼ 1 f ðmodeÞÞ−Minð∑i ¼ 1 f ðmodeÞÞ
Inter rater agreement ð%Þ ¼
i ¼ 1; ::; 10 features j ¼ 1; ::; n subjects
ð1Þ
The suggested benchmarks for reliability are as follows: poor to slight, 0–20%; fair, 21–40%; moderate, 41–60%; substantial, 61–80%; and excellent, 81–100% [13].
2.4.2.
Left and right lung asynchrony
The vibrational energy in dB can be plotted over time for each lung, with time on the x-axis and dB on the y-axis (Fig. 1). The timing of the inspiratory and expiratory VEPs was compared for the left and right lungs [6]. The gap index was defined as the absolute value of the average of the gaps (s) between the VEPs of both lungs for 12 s.
2.4.3.
Regional lung asynchrony
Although whole left lung–right lung asynchrony was confirmed by the gap index, regional left lung–right lung asynchrony might have been masked if there were differences in the time course of regional airflow. The time course of vibrational energy for each row (6–7 rows for each lung), in addition to that for the whole lung, can be plotted and used to analyze regional lung asynchrony (Fig. 2). Because VRI collects data 71 times in 12 s, the recording period can be divided into 70 sections of 0.17 s. We defined the asynchronous section in each row as the section in which the change (increase or decrease) in vibrational energy of the right and left rows was discordant. The asynchrony score is defined as the average number of asynchronous sections in each row.
2.5.
Statistical analysis
A p-value of o0.05 was considered significant for all tests. Significant differences in the regional lung QLD between the left and right lungs were evaluated using two-sided t-tests for paired data. Differences in the regional QLD, the gap index, and the asynchrony score between male and female subjects were also evaluated using two-sided t-tests for unpaired data. The qualitative assessments from the raters were analyzed with comparisons between the male and female subjects and between the non-smokers and smokers by using chi-squared tests.
31
respiratory investigation 52 (2014) 28 –35
Gap between vibration energy peaks (VEP) of lungs Left
Time course of vibration energy(frame)
Right
Vibration energy (dB)
-35
Left lung
Right lung
-40
VEP gap -45 -50 -55 -60 -65
inspiration
expiration
the gap index = |average of VEP gap (inspiratory and expiratory) | Fig. 1 – Gaps between the vibrational energy peaks (VEP) of the lungs. The vibrational energy in dB can be plotted over time for each lung, with time on the x-axis and dB on the y-axis. The timing of the inspiratory and expiratory VEPs was compared for the left and right lungs. The gap index was defined as the absolute value of the average of the gaps (s) between the VEPs of both lungs for a duration of 12 s.
Regional left and right lung asynchrony Time course of vibration energy In bilateral 1st row (frame) Right 1st row
•• •• ••
• • •
-40
Vibration energy (dB)
Left 1st row
-45 -50 -55 -60 -65 -70 -75 Lt 1st row
Rt 1st row
●●
●
the asynchrony score (AS) = average number of the asynchronous sections of each row
●: asynchronous section Fig. 2 – Regional left and right lung asynchrony. The time course of vibrational energy for each row can be plotted and used to analyze regional lung asynchrony. Because VRI collects data 71 times during 12 s, the recording period can be divided into 70 sections of 0.17 s. We defined the asynchronous section in each row as the section at which the change (increase or decrease) in vibrational energy of the left and right rows was discordant. The asynchrony score is defined as the average number of asynchronous sections for each row.
3.
Results
3.1.
Characteristics of the subjects
Recordings were made of 57 male and 56 female subjects. One female and 2 male subjects were excluded during the screening stage for not meeting the eligibility criteria (1 with asthma and 2 diagnosed with restrictive lung disease after the lung
function tests). One male subject was excluded because of missing lung function test results, 2 male subjects were excluded because of missing inter-rater agreement data, and 1 female subject was excluded because of image artifacts. The remaining 106 subjects (52 male and 54 female; mean age, 33.779.6 years), including 67 non-smokers and 39 smokers (11 ex-smokers and 28 current smokers), were included in this study.
32
respiratory investigation 52 (2014) 28 –35
The demographics, anthropometric values, and lung function test results of the study population are shown in Table 1. The smoking rates in female subjects were significantly lower than those in male subjects (po0.005). Male smokers were significantly older than male non-smokers (po0.05). Male smokers had significantly lower forced expiratory volume in 1 s (FEV1)/forced vital capacity (FVC) compared with male non-smokers (po0.05).
3.2.
QLD analysis
The average total left QLD value, including the upper, middle, and lower QLD values, was significantly higher than the corresponding value in the right lung for all subjects, with the exception of the lower regions in male subjects (Table 2). The dominance of left lung QLD was more prominent in female subjects. The average right upper, middle, and total QLD values were significantly lower, whereas the average left lower and total QLD values were significantly higher in female subjects. Subgroup analysis in non-smokers showed that the dominance of the total left lung QLD was significant in all subjects (Table 2). However, there were no significant differences between the total left and right lung QLD values in smokers (Table 2).
3.3.
Qualitative assessment of VRI images
The overall inter-rater agreement for all features was 82.9%. For the final assessment, 81.1% of all subjects were considered normal. The ratios for good dynamic image, good image development, good MEF shape, symmetric MEF area, symmetric MEF intensity, and no missing parts were 87.7%, 74.5%, 85.8%, 84.9%, 78.3%, and 91.5%, respectively. The final assessments and all parameter scores for the images, except for the dynamic images, were significantly lower in female subjects (Fig. 3). We analyzed inspiratory peaks in 106 subjects and expiratory peaks in 102 subjects because expiratory peaks were not distinct for 2 male non-smokers and 2 female non-smokers. The average gap index of the 106 subjects was 0.0670.05 s,
and the gap index was lower than 0.06 s in 76 subjects (71.7%). The average asynchrony score for all subjects was 10.873.5. Although no significant difference between the sexes was observed in the gap index (female subjects, 0.0670.06 s; male subjects, 0.0570.04 s), there was a significant difference in the asynchrony score between female and male subjects (female subjects, 12.573.6; male subjects, 9.172.4; po0.001, Fig. 4). We did not find any significant differences in the qualitative assessments between the non-smokers and the smokers (Table 3).
4.
Discussion
Our QLD results were similar to those of a previously published report [4]. However, there were discrepancies in the effects of sex on the qualitative assessment. A significantly greater number of female subjects in our study had abnormal images, as judged by the raters. The mathematical analyses we introduced for VRI assessment also revealed significantly higher regional lung asynchrony in female subjects. Because this phenomenon was not found in a previous study of Israeli subjects [4], the VRI variability in sex demonstrated in this study may be considered as a normal characteristic among the Japanese population. VRI is a commercially developed acoustic lung imaging system that displays breath sound distributions as dynamic gray-scale images [2,3]. Reports in which VRI was applied in clinical settings, such as in interventional bronchoscopy, lung transplantation, pleural effusion, pneumothorax, foreign body aspiration, and obstructive lung diseases, have been published [6–9,14–18]. Yigla [4] reported the characteristic features of VRI in healthy asymptomatic subjects; the researchers reported that 89% of the healthy population had a normal final assessment. The total left lung QLD value was significantly higher than the right lung QLD value. This left lung dominance in QLD was more prominent in female subjects and was weaker in smokers. Although the subgroup analysis did not show any significant effects of sex using the qualitative assessment, a significantly higher proportion of heavy smokers had images rated as abnormal.
Table 1 – Characteristics of subjects. Male
No. of subjects Brinkman index Pack-years Age (y) Height (cm) Weight (kg) BMI FVC (% predicted) FEV1 (% predicted) FEV1/FVC (%)
Female
All
Non-smokers
Smokers
All
Non-smokers
Smokers
52 151.57217.8 7.6711.0 34.379.1 171.975.2 69.679.5 23.572.8 99.878.6 95.378.2 83.975.3
25 0 0 30.977.4 171.875.0 68.6710.3 23.273.0 97.278.1 94.677.2 86.275.4
27 303.07221.5 15.2711.1 36.178.7* 171.975.4 70.778.8 23.972.7 102.178.8 96.079.2 82.374.3**
54 30.4776.4 1.573.8 33.8710.7 158.375.1 53.377.0 21.372.8 102.9711.8 103.8711.3 87.475.9
42 0 0 34.2711.7 157.875.0 52.476.9 21.072.4 102.8712.8 103.9711.1 87.775.8
12 136.87110.9 6.875.5 32.175.8 159.875.7 56.376.5 22.274.0 103.478.0 103.4712.5 86.576.3
Definition of abbreviation: BMI ¼body mass index, FVC ¼forced vital capacity and FEV1 ¼ forced expiratory in 1 s. Values are represented as means7standard deviation. n po0.05. nn po0.01. Compared with non-smoker.
33
6.773.4 14.575.3 24.575.8 45.779.0 8.972.1# 20.876.1# 24.677.2 54.379.0
Smokers n ¼12
respiratory investigation 52 (2014) 28 –35
(%) 100
Qualitative assessment of VRI images *
80 60
*
*
* *
*
5.973.2 13.373.7 21.475.5 40.677.0 10.272.6### 22.874.6### 26.574.8### 59.477.0###
20
6.173.2ǁǁ 13.674.1ǁǁǁ 22.175.7 41.777.7ǁǁǁ 9.972.5### 22.375.0### 26.175.4#ǁǁǁ 58.377.7###ǁǁǁ
0
Male (n=52) Female (n=54)
*p
