The study was designed as a prospective, observational study. The first part of the study compared the new method with the two previously described ultrasound methods to assess diaphragm movement, using exhaled air volume as reference test. The second part evaluated agreement between two expert ultrasound operators. The third part examined feasibility of the different methods performed by novice ultrasound operators.
First part
Participants
Healthy volunteers were eligible for inclusion after giving oral informed consent. Exclusion criteria were diaphragm dysfunction, any neuromuscular, neurological, or pulmonary disease. Nineteen individuals were enrolled.
Breathing pattern and participant positioning
Participants were placed in erect position, allowing simultaneous spirometry measurement and ultrasound recording of film clips and images. Participants performed several breathing maneuvers that all were recorded and analyzed. Participants were instructed to inhale and exhale slowly to allow simultaneous ultrasound examination during the entire breathing cycle. Forced exhalation was avoided not to displace the ultrasound transducer. Measurements were performed with different breathing patterns decided by the participants, ranging from maximal inspiration to shallow breathing.
Spirometry measurement
All participants exhaled through a Vitalograph, Spirotrac (Hamburg, DE) spirometer. The spirometer was calibrated before analysis following manufacturer’s instruction. Volume of exhaled air was measured with the spirometer in a “slow vital capacity” modus. In this way, total volume was quantified.
Ultrasound measurement
A single experienced ultrasound operator performed all ultrasound examinations. All ultrasound examinations were done simultaneously with spirometry recordings of exhaled volume. The ultrasound operator was blinded to spirometry measurement results. Film clips were saved for later analysis. A curvilinear 3–5 MHz probe and a General Electric Vivid S8 (Little Chantfort, UK) ultrasound machine with a standard abdominal preset were used for ultrasound measurements. Analysis of diaphragm movement performed on ultrasound films and images was done blinded to volume of exhaled air measured by spirometry. The following three different methods were performed in all participants.
M-mode measurement
The M-mode measurement has been described in several studies [11,12,13,14]. Right hemidiaphragm was visualized using a subcostal view in the mid-clavicular line with the probe tilted cranially. The liver served as an acoustic window to identify the posterior part of the diaphragm. A M-mode line was placed over the posterior part of the diaphragm with maximal movement, and the excursion was measured in millimeters during breathing (Fig. 1).
B-mode measurement
The B-mode was performed in accordance with Gethin-Jones et al. [15]. In a mid-axillary view, the cranial top of right hemidiaphragm was visualized. A film clip was saved during respiration, and maximal cranial–caudal movement of the diaphragm top was measured in millimeters (Fig. 2).
Area method
Changes in the intra-thoracic area during respiration were calculated using the Area method, which is described in detail here. A further instruction on how the Area method is performed is available in the Additional file 1: Video 1 and Additional file 2: Video 2. The same film clip of the respiration as used to in the B-mode measurement, a recording of the diaphragm movement from a mid-axillary lateral view, was used. At the right side, the liver was used as a sonographic landmark to identify the right hemidiaphragm. At the left side, the spleen was the landmark. Scrolling through the film an image frame with maximal diaphragm contraction, corresponding to end-inspiration, was identified. With the ultrasound machine’s build-in area-calculation function, the entire visible portion of the diaphragm was traced (Fig. 3). In case of limited view to a part of the diaphragm, tracing continued the curve obtained from visible parts. The ultrasound transducer was kept in a fixed position during the respiratory maneuver, allowing the borders of the ultrasound image to be used as limits of the area. These limits did not change during the respiratory maneuver. The only change in area was due to the change in position of the diaphragm with respiration. In this manner, the intra-thoracic area was calculated in maximal diaphragm contraction. Then, scrolling through the film, the frame with minimal diaphragm contraction, corresponding to pre-inspiration, was identified. The intra-thoracic area was calculated by tracking the diaphragm in the same way (Fig. 4). Subtraction of the area with maximal diaphragm contraction from the area with minimal diaphragm contraction gave the change in intra-thoracic area during the breathing maneuver: Δ intra-thoracic area during respiration = intra-thoracic area in maximal diaphragm contraction − intra-thoracic area in minimal diaphragm contraction.
Second part
Inter-rater variability
To evaluate the inter-rater variability, two independent observers (SHS, CBL) viewed 40 ultrasound film clips. Observers analyzed film clips and measured diaphragm movement using all three methods. The analysis was blinded to the other observer and blinded to volume of exhaled air.
Third part
Feasibility
The feasibility to obtain ultrasound images of the diaphragm in the different scanning positions was studied in healthy volunteers. Novice ultrasound operators completed a 1-day course in point-of-care ultrasonography and were introduced to ultrasound examination of the diaphragm in a 1-h practical training session.
Scanning positions and imaging technique
The novice operators were asked to produce ultrasound images of both hemidiaphragms needed to make the different measurements. Ultrasound examination of the right and left mid-clavicular lines was performed to do the M-mode measurements. Ultrasound examination of the right and left mid-axillary lines was performed to do the B-mode and Area measurements.
Measurement technique
The novice ultrasound operators were asked to do M-mode and Area measurements on stored images or film clips, following instructions of an expert operator. They were asked to rate difficulty of each method on a 5-point Likert scale, where 1 was given if the measurement was very easy to perform, 2 if it was easy, 3 if it was moderately difficult, 4 if it was difficult, and 5 if it was very difficult to perform.
Statistics
Statistics were performed using Stata vs.14.2 (StataCorp, TX, USA). For continuous data, quantile–quantile plots were done to assess parametric distribution. Spearman’s ranked-order correlation coefficient (rs) was used to analyze associations between non-parametric variables. Agreement between raters was calculated with one-way intra-class correlation. Binominal data were analyzed using exact methods. Estimates are presented with 95% confidence intervals. Significance level was set to equal p ≤ 0.05. Graphs were made using GraphPad Prism (GraphPad Software, CA, USA). No prior studies could be used to provide data for sample size calculations.