Lung volume reproducibility under ABC control and self‐sustained breath‐holding

Abstract An Active Breathing Coordinator (ABC) can be employed to induce breath‐holds during CT imaging and radiotherapy of lung, breast and liver cancer, and recently during lung cancer MRI. The apparatus measures and controls respiratory volume, hence subject lung volume reproducibility is its principal measure of effectiveness. To assess ABC control quality, the intra‐session reproducibility of ABC‐induced lung volumes was evaluated and compared with that reached by applying the clinical standard of operator‐guided self‐sustained breath‐holds on healthy volunteers during MRI. Inter‐session reproducibility was investigated by repeating ABC‐controlled breath‐holds on a second visit. Additionally, lung volume agreement with ABC devices used with different imaging modalities in the same institution (MR, CT), or for a breast trial treatment, was assessed. Lung volumes were derived from three‐dimensional (3D) T1‐weighted MRI datasets by three observers employing semiautomatic lung delineation on a radiotherapy treatment planning system. Inter‐observer variability was less than 6% of the delineated lung volumes. Lung volume agreement between the different conditions over all subjects was investigated using descriptive statistics. The ABC equipment dedicated for MR application exhibited good intra‐session and inter‐session lung volume reproducibility (1.8% and 3% lung volume variability on average, respectively). MR‐assessed lung volumes were similar using different ABC equipment dedicated to MR, CT, or breast radiotherapy. Overall, lung volumes controlled by the same or different ABC devices agreed better than with self‐controlled breath‐holds, as suggested by the average ABC variation of 1.8% of the measured lung volumes (99 mL), compared to the 4.1% (226 mL) variability observed on average with self‐sustained breath‐holding.


| INTRODUCTION
Imaging and radiotherapy of the thorax and abdomen are adversely affected by respiratory motion. Breath-holding for short time intervals is a widely applied technique to reduce this effect. In standard clinical practice, patients are instructed to hold their breath, 1 yet the true onset, constancy and reproducibility of individual self-sustained respiration control is questionable, especially in the absence of respiratory monitoring, and depends on patient compliance. To address this issue, a volumetric respiratory monitoring and control apparatus that induces breath-holds automatically at a predefined inhaled or exhaled air volume during a preset duration (active breathing coordinator, ABC) has been developed by Elekta (Elekta Oncology Systems Ltd, Crawley, West Sussex, UK). The ABC can be employed during lung, 2 breast 3 and liver radiotherapy 4 to minimize breathing motion and consequently to reduce treatment margins and/or to reduce dose to healthy tissue. An ABC consists of a breathing tube with a mouthpiece and filter, connected to a digital volume transducer & pickup assembly and a balloon valve that can be inflated to halt respiration. 5 The goal of the device is to induce reproducible breathholds at the same lung volume, during and across sessions, whether this may be radiotherapy delivery or CT or PET examinations. Lung volume reproducibility is therefore the main characteristic of the ABC respiratory control effectiveness.
In addition to CT, PET, and radiotherapy, ABC has been recently used in an MR setting. 6 8 As ABC control offers the potential to produce wellmatched images during different acquisition sessions, it is important to assess lung volume consistency between such sessions. In clinical practice, similar ABC devices of the same model are dedicated to different imaging or treatment modalities, thus air volume repeatability should also be compared between these devices.
In an ABC feasibility study, McNair et al. 9 have observed that outlined lung volumes on planning CT scans agreed with tidal volumes recorded by the ABC, but that the ABC-displayed breath-held volumes were affected by flow rate. As this effect may raise concerns about the reproducibility of the air volumes controlled by the ABC, actual lung volume measurements on human subjects are necessary for clarification. Treating breast cancer patients, Bartlett et al. 10 found no significant differences in lung volumes between ABC control and a specific voluntary breath-holding technique where breath-hold consistency was monitored by the position of patient tattoos relative to lasers. Using a different spirometer, Fassi et al. 11 found that spirometer-based control did not guarantee a reproducible position of the external breast surface in deep inhalation breath-hold (DIBH) left-breast radiotherapy. Nevertheless, measuring whole lung volumes from three-dimensional (3D) imaging would present a more accurate evaluation of spirometer-based respiration control than external markers. Hunjan et al. 12 have also observed that abdominal external fiducial extrema positions differed between breath-holding and free breathing.
Single organs of interest have primarily been assessed during radiotherapy: Eccles et al. 4 have observed a good intra but less satisfactory inter-fraction reproducibility of liver position using ABC; employing a different active breathing control device for breast radiation therapy, Moran et al. 13 15 have shown a 2.5 times higher inter breath-hold than intra-breath-hold variation by measuring the reproducibility of diaphragm position during verbally coached voluntary DIBH maneuvers.
The present work serves as a quality assurance of clinically used ABC devices during clinically applicable imaging of human subjects without radiation exposure. We evaluated intra-session and intersession lung volume reproducibility achieved with ABC control, by repeating breath-holds with the modified MR-compatible ABC system applied to lung cancer imaging. 6 The study was performed on healthy volunteers, assuming no significant physiological changes in the course of one to 4 weeks between two imaging sessions. Moreover, we investigated if similar ABC devices used with different clinical imaging modalities (MR, CT, breast trial radiotherapy) reliably provide the same lung volumes. In addition, we examined whether ABC provides more reproducible lung volumes than the clinical standard of self-sustained breath-holding, where an operator instructed each volunteer to perform repeated self-controlled breath-holds. In all cases, lung volume measurements were extracted from 3D T 1weighted volumetric MR images acquired during each breath-hold.

2.A | ABC set up
We used the modified MR-compatible ABC system described by Kaza    The VIBE measurement was repeated four times in total with the MR-ABC kit for each volunteer, to assess intra-session lung volume reproducibility with this kit. The measurement was repeated four times with the CT-ABC and trial-ABC, in order to assess intrasession lung volume reproducibility with each of these kits and to investigate possible kit-related variations. The ABC kits were exchanged by an operator as described in section A, without moving the volunteer. For comparison with self-induced suspension of respiration, the VIBE measurement was also acquired four times in selfcontrolled breath-holds using the ABC spirometer. The scanner was then prepared as before and subjects were instructed to "breathe in, breathe out, breathe in, breathe out, breathe in and hold" as in com-     Table 1 for each of the two batches comprising four 3D image datasets. The mean absolute bias between observers 1 and 2 (0.5%) is smaller than the bias between observers 1 and 3 (2.7%) and observers 2 and 3 (3.0%). The average absolute bias and reference interval from all observers and image datasets was 2.1 AE 3.7%. Table 2 demonstrates the mean and standard deviation of the delineated volumes from the four 3D image datasets with the same breath-holding condition for each subject. Kit intra-session variability of measured lung volumes was similar for the two MR-ABC sessions (98 mL and 97 mL), and comparable to that for CT-ABC and trial-ABC (112 mL and 91 mL, respectively). The average MR-ABC kit intra-session variability from the two visits was 1.8% of the delineated lung volumes. The overall ABC intra-session lung volume variation, as expressed by the average standard deviation of all ABC conditions over all volunteers, was 99 mL, representing 1.8% of the mean lung volume under ABC control. In contrast, the average standard deviation of the self-sustained breath-holds over subjects (self-sustained intra-session lung volume variation) amounted to 4.1% of the mean self-controlled volume (226 mL).

3.B | Lung volume reproducibility of breath-holding conditions
In general, volume standard deviation ranged from 33 to 185 mL under ABC control, but from 105 to 594 mL for self-induced respiration holds. The absolute lung volume difference between the two MR-ABC sessions was 3% on average, suggesting good intersession lung volume reproducibility. The CT-ABC and trial-ABC volumes differed by 2 and 5% on average from MR-ABC session 1, also indicating good lung volume reproducibility between the three ABC kits.

| DISCUSSION
In this work, we took advantage of the recently available MR-com-    Italics indicate the percent difference of the mean lung volume of each ABC specific breath-holding condition from the mean lung volume of MR-ABC session 1. n.d.: not defined.

(a) (b)
F I G 2 . (a) A partition of the 3D image datasets of BH1 and BH4 of condition MR-ABC 2 for Volunteer 5, with a small difference (4 mL) between their delineated lung volumes. Images are displayed in the acquired axial and reconstructed coronal and sagittal orientation. The image dataset of BH4 was rigidly registered on the BH1 dataset. The solid red and dotted yellow lines represent the delineated lung volume of BH1 and BH4, respectively. The images and lung contours demonstrate an excellent agreement. (b) A similarly positioned partition of selfsustained BH4 rigidly registered on self-sustained BH2 for the same volunteer, with a high difference (1370 mL) between the two delineated lung volumes, displayed in all three orientations. The images of the two breath-holds present notable discrepancies in diaphragm, vessels and thorax position, displayed as blue and orange shaded regions.
available within the widely used TPS RayStation. The large amount of acquired MR data (120 volumetric image datasets) was divided equally between three observers for lung delineation. For consistency, all 20-image datasets of the same subject were processed by the same observer. Inter-observer variability was estimated as a few percent of the calculated lung volume, using the four datasets of a volunteer and breath-holding condition with typical image quality, and those of a case presenting imaging artifacts. Bland-Altman analysis indicated that bias between observers may reach 4% while variation may be as much as 6%. The smaller bias between Observers 1 and 2 suggests that they followed a more similar lung contouring procedure than Observer 3. Image quality appeared to have a smaller impact on variability than the inevitable subjectivity of manual editing around ambiguous structures such as vessels and the diaphragm, which the automated algorithm cannot accurately contour.
This work presents the raw data of the calculated lung volumes from all subjects and breath-holding conditions, and summary statistics. Performing four breath-holds allowed the variability between breath-holds to be assessed for each condition, and averaging these volumes improved the accuracy of comparisons between the different conditions. The total number of performed breath-holds in one session should be restricted to avoid volunteer fatigue, and reached 16 in our study. The duration of the induced breath-holds was close to the clinical limit of 20 s. Given the small number of subjects we implemented descriptive statistics only, avoiding hypothesis testing.
Our data suggest that the ABC-controlled breath-holding volumes were more closely clustered than the self-sustained ones. The overall ABC intra-session lung volume variability (1.8% of the mean lung volume attained using ABC), was less than half the self-sustained intra-session lung volume variability. However, one subject (Volunteer 5) presented the largest standard deviation for self-