Actuated Dose Delivery and Physiological Monitoring of ¹²⁹Xe MRI: Implications for Improving Repeatability

Abstract

Objective:

Identify and quantify physiological sources of variability in hyperpolarized ¹²⁹Xe gas-exchange MRI and present a dose-delivery workflow that promotes repeatable lung inflation and alveolar pressure during imaging.

Theory and Methods:

A modular ¹²⁹Xe dose-delivery and monitoring system was developed with two configurations: a remotely actuated flow/volume-monitoring device used to evaluate pre-dose lung inflation, and a pressure-sensing mouthpiece used with conventional delivery to estimate alveolar pressure during breath-hold. In Cohort A (n=12), standard coaching was evaluated by quantifying how pre-dose lung volumes deviated from the target of functional residual capacity (FRC). In Cohort B (n=14), ¹²⁹Xe spectroscopy under normal, Mueller, and Valsalva breath-hold maneuvers was used to quantify the extent to which alveolar pressure modulates RBC:Membrane (RBC:M) and RBC oscillation amplitude.

Results:

In Cohort A, coaching drove subjects in 70% of tests to exhale below FRC prior to dose delivery. Reduced lung volume is known to increase membrane and RBC signal beyond healthy-reference ranges. In Cohort B, changing alveolar pressure inversely affected RBC:M; Mueller increased RBC:M by 4.4% and Valsalva decreased it by 7%. Both maneuvers reduced RBC oscillation amplitude by ~20–25% of the original amplitude.

Conclusion:

Variations in lung inflation and alveolar pressure are significant drivers of variability in critical ¹²⁹Xe metrics. Integrating actuated delivery and real-time monitoring into dose delivery workflows offers a practical pathway to improving repeatability and standardization of quantitative spectroscopy and gas-exchange imaging.

Introduction

Hyperpolarized ¹²⁹Xe MRI is a rapidly developing technology that enables comprehensive breath-hold, non-invasive 3D imaging of pulmonary function. This technique includes not only quantitative imaging of ventilation, which is now FDA approved [1], but also analysis of gas exchange from the airspaces to the alveolar membrane tissues and red blood cells (RBCs) [2–4]. Thus, the ability of ¹²⁹Xe MRI to trace the same pathway as oxygen has proven useful in characterizing obstructive, restrictive, and pulmonary vascular disease [5–8]. In addition to these static gas-exchange measures, the dissolved-phase ¹²⁹Xe signal arising from RBCs oscillates at the cardiac cycle, reflecting dynamic variations in pulmonary capillary blood volume that speak to the presence of pulmonary hypertension [9]. Furthermore, the procedure is non-invasive and involves no ionizing radiation, permitting patients and healthy volunteers to safely undergo ¹²⁹Xe MRI repeatedly [10–12]. Therefore, ¹²⁹Xe MRI is positioned to not only guide treatment decisions but to track health progression and therapy response over time [13].

While ¹²⁹Xe imaging technology and applications have advanced considerably over the past decade, delivery of hyperpolarized ¹²⁹Xe gas to the patient is relatively unchanged since the first human studies nearly 3 decades ago [14]. Typically, hyperpolarized ¹²⁹Xe gas is dispensed from the polarizer into a dose delivery bag made of polyvinyl fluoride, a material chosen for its relatively benign effect on the relaxation of spin-polarized ¹²⁹Xe [15]. The bag is then inflated to capacity with inert buffer gas and clipped shut on the attached Tygon tube. Most studies, as well as the FDA-cleared product, then fit the tubing with a mouthpiece for delivery. In the early stages of the field, all patients received gas from a 1-liter bag. However, subsequent work began to observe the utility of tailoring the total volume to either 1/6 of the patient’s total lung capacity (TLC) [16] or to 1/5 of their forced vital capacity (FVC) [17]. Likewise, most studies now strive to have patients begin their dose inhalation from functional residual capacity (FRC) as it has been found to improve repeatability [18]. However, current procedures cannot guarantee this target is met because there are no existing methods to verify lung inflation during dose delivery. Consequently, the result of each session depends on the coaching provided by study personnel and the performance of the patient. Additionally, the lack of coaching standardization of ¹²⁹Xe MRI brings complexity to data interoperability and image repeatability between sites.

Given that ¹²⁹Xe dose administration is central to the success of the imaging session, analysis of the delivery workflow is worthy of specific attention. One practical limitation of administration is that the delivery of the gas requires precise timing; the patient must complete preparatory breaths after which the study personnel must quickly place the dose mouthpiece in the patient’s mouth, unclip the bag, and finally instruct the patient to inhale its contents [18]. This, in turn, requires a dedicated team member to perform the dose administration while the MR technologist focuses on synchronizing image acquisition with the gas delivery. A more streamlined solution would allow the MR technologist to coach the patient and administer the ¹²⁹Xe dose while also initiating the scan, similar to any typical contrast media study.

Beyond inefficient logistics, the current dose delivery practices have the potential to drive variability in the quantitative metrics derived from the scan. Such variability has repeatedly been shown to arise from scan-to-scan differences in lung inflation [19]. This particularly impacts the signal ratios Membrane:Gas and RBC:Gas because changes in lung tissue density and physiology at different inflation volumes alter pulmonary gas exchange. As the starting lung inflation at dose delivery deviates from FRC, repeatability of these ¹²⁹Xe MRI-derived metrics worsen. Given the emerging evidence of metrics like Membrane:Gas being a predictor of antifibrotic response in IPF [20], it becomes paramount to measure it with the highest possible repeatability. However, with current approaches, if a healthy individual were imaged at a lung volume below their FRC, the resulting scans would exhibit high Membrane:Gas uptake resembling patterns normally seen in ILD [8,21]. Similarly, a healthy individual imaged at a lung volume close to their TLC would exhibit low membrane uptake, normally associated with COPD [19]. While such issues are somewhat mitigated by the use of tailored dose volumes, these improvements can still be undermined by variability in the starting lung volume before dose inhalation.

A final factor affecting imaging physiology is the alveolar pressure during the breath-hold. This effect has received scant attention in the field, although prior work has suggested that patient actions during the breath-hold, such as Valsalva and Mueller maneuvers, can measurably change the RBC:Membrane ratio (RBC:M) [22,23], a measure of the partitioning of ¹²⁹Xe signal between RBCs vs lung tissue and plasma [24]. Valsalva maneuvers involve attempting to exhale against a closed glottis, thereby increasing alveolar pressure; conversely, Mueller maneuvers are attempts to inhale in the same scenario, thereby decreasing alveolar pressure. Both maneuvers can be performed by participants as they hold their breath during scanning, often inadvertently. Since these maneuvers affect alveolar pressure, they are expected to change pulmonary capillary recruitment/distension and thus capillary blood volume [25]. When alveolar pressure rises above that of the capillaries, their blood volume is reduced, as is ¹²⁹Xe-RBC transfer. By contrast, a reduced alveolar pressure leads to capillary recruitment and distension and thus increases ¹²⁹Xe-RBC transfer. Such pressure-induced changes in recruitment are also expected to change the amplitude of cardiogenic RBC signal oscillations, which are known to be modulated by capillary blood volume [26].

Here, we introduce a potential solution that addresses three limitations of current 129Xe MRI workflows: dose delivery, lung inflation repeatability, and alveolar pressure variability during imaging. We developed a modular ¹²⁹Xe dose delivery and physiological monitoring system with two configurations. The first used a remotely actuated device that enabled manual dose delivery from outside the scanner bore while providing real-time monitoring of respiratory flow and volume. The second configuration used a separate pressure-sensing mouthpiece with conventional dose delivery to estimate alveolar pressure during the acquisition breath-hold. We present the results of initial studies using this system to evaluate dose delivery performance and to assess the effects of lung volumes and breath-hold pressure maneuvers on ¹²⁹Xe gas-exchange metrics. Ultimately, this work may help standardize ¹²⁹Xe delivery procedures across sites, thus guiding patients to be imaged with repeatable lung inflation volumes.

Design and Theory

Remote ¹²⁹Xe Dose Administration Device

The remote dose administration device is shown in Figure 1. It was designed to attach to the standard dose delivery bag by replacing the current mouthpiece. After it is affixed to the bag, the retaining clip can be released as the device now governs gas outflow; the mouthpiece is then placed in the participant’s mouth prior to actuated delivery. In this state, the device blocks the flow of ¹²⁹Xe from the bag, allowing patients to breathe room air. At the time of the scan, the MRI tech can initiate ¹²⁹Xe dose delivery outside of the scanner bore by pneumatically activating the internal piston, thereby allowing the participant to inhale the dose from the bag. The device was designed for injection molding and constructed with medical-grade polymers; its main body, cap, and mouthpiece were manufactured with Dow HDPE DMDA 8907 NT7 while the piston consisted of Versaflex CL2250 TPE. The pneumatic activation is controlled by an air-filled syringe connected to the switch piston cavity via Tygon tubing and Luer Lock connectors.

Figure 1.

A cross-sectional view of the remote ¹²⁹Xe delivery device. A Venturi tube is appended below the dose inlet to allow the high and low pressures to be measured by a differential pressure sensor mounted on the MRI table. The piston starts in the lowered position for room air breathing and is raised for dose delivery by a pressure pulse from a syringe connected to the pressure inlet/outlet. All pressure outlets utilize Luer fittings to quickly connect tubing to external devices.

Modifications for Inspiratory and Expiratory Volume Monitoring

By appending a Venturi tube between the piston switch and the mouthpiece, the pneumatic dose delivery device was modified to also monitor respiratory flows and volumes prior to ¹²⁹Xe inhalation. With this addition, volume flow through the device can be calculated by measuring the differential pressure between the high- and low-pressure zones of the Venturi tube. Using Bernoulli’s principle, the measured pressure differential was converted to flow velocity and subsequently to volumetric flow rate, which was then integrated to determine the volume of air displaced during respiration. The Venturi connections were connected via Tygon tubing to a differential pressure sensor (SparkFun SDP31) enclosed in 3D printed polylactic acid (PLA) housing. The sensor was connected by a 15’ RJ11-style cable to a microcontroller (ESP32 WROVER) housed in a separate 3D printed enclosure secured to the table railing outside the scanner’s fringe field (less than 10mT). The controller was used to provide real-time data visualization by calculating the volume flow of each breath through the device. This data was also stored locally and then exported for more detailed analysis. The full experimental setup for this device is shown in Figure 2.

Figure 2.

Device setup for ¹²⁹Xe dose delivery with respiration monitoring. The remote delivery device with dose bag is placed in the subject’s mouth. A nose clip is applied, if necessary, to prevent nasal breathing. The subject is able to breathe ambient air until the MRI personnel initiate dose delivery by depressing the syringe connected to the pressure inlet/outlet of the device.

Alveolar Pressure Estimation Device

During inspiration, contraction of the diaphragm expands the thoracic cavity, lowering alveolar pressure below that at the mouth and causes air to flow into the lungs. Between breaths, alveolar pressure equalizes to that at the airway opening and causes airflow to cease. During expiration, relaxation of the diaphragm increases alveolar pressure and causes air to be expelled from the lungs. While alveolar pressure cannot be directly measured non-invasively, during periods when airflow is stopped, it can be assumed to be equal to the pressure at the mouth [27]. Thus, as a surrogate for measuring alveolar pressure, a separate device was designed to measure oral air pressure during the dose delivery and subsequent breath-hold as seen in Figure 3. At the time of the scan, the MRI tech can perform their coaching without the device and then transition to traditional ¹²⁹Xe delivery by placing the device in the subject’s mouth and unclipping the dose bag. Upon full dose inhalation, the device remains in the subject’s mouth and occludes airflow, allowing mouth pressure measured throughout the scan to be used as an estimate of alveolar pressure. To measure positive pressures while preventing airflow, a check valve was incorporated between the dose bag and the pressure sensor.

Figure 3.

A sectional analysis of the pressure-sensing dose delivery device. The ¹²⁹Xe dose bag is attached to the top of the device and not unclipped until ready for delivery. Upon delivery, the check valve is pulled down by the subject’s inspiration, which can be continued into a Mueller maneuver. When the subject exhales, the check valve moves up to prevent gas reentering the bag, enabling the Valsalva maneuver. The device is connected via Tygon tubing with Luer fittings to a sensor used to estimate alveolar pressure from mouth pressure during breath-hold. The mouthpiece is injection molded and separate from the device.

The main body of the device was 3D printed and constructed out of medical grade Polyethylene Terephthalate Glycol (PETG). The check valve was injection molded with platinum-catalyzed addition-cure polysiloxane elastomer (RTV-2). The mouthpiece was injection molded with Dow HDPE DMDA 8907 NT7. The device’s pressure outlet was connected by Tygon tubing and Luer Lock connectors to the pressure sensor (HX710B) contained in a 3D printed (PLA) housing. The sensor was connected by a 15’ RJ11-style cable to a microcontroller (ESP32 WROVER model) enclosed in 3D printed housing secured to the table railing outside the scanner’s fringe field (less than 10mT). The controller was used to provide real-time alveolar pressure estimations to the user, which was stored locally and later exported for more detailed analysis. The full experimental setup for this device is shown in Figure 4.

Figure 4.

Device setup for continuous pressure monitoring. The pressure-sensing delivery mouthpiece with dose bag is held by MRI personnel until the patient is ready for administration. The subject is in the bore and has a nose clip applied, if necessary, to prevent nasal breathing. Upon inhaling the dose, the subject is asked to perform either a normal breath-hold, a Mueller, or a Valsalva maneuver during the scan.

Methods

Study Participants

Two cohorts of healthy 18–69-year-old participants were recruited consecutively for separate device-testing protocols: Cohort A (n=12) participated in the respiration monitoring with actuated dose delivery study, and Cohort B (n=14) participated in the alveolar pressure sensing with traditional dose delivery study. Both cohorts had approximately equal sex distribution and none had prior ¹²⁹Xe MRI experience. None had a history of pulmonary disease, all had less than five pack-years of smoking history with no smoking history within the last five years, and no history of using other inhaled products more than once per week for over one year. All participants were studied at Duke University Medical Center under harmonized HIPAA-compliant IRB protocols and hyperpolarized ¹²⁹Xe doses were administered under Duke’s investigational new drug application 109,490.

Respiration Monitoring Prior to and During ¹²⁹Xe Delivery

In Cohort A, the remotely actuated flow/volume-monitoring configuration was used to measure respiratory flow and lung volume changes as participants underwent conventional breath coaching and inhaled the contents of the dose bag. For each individual dose, prior to starting measurements, the remote delivery device was calibrated to account for environmental temperature and pressure differences along the respiration flow direction in its Venturi tube. Calibration was necessary as reverse flow through a Venturi tube induces turbulence as gas travels though the low-pressure constriction to the high-pressure dilation, requiring a calibration coefficient to normalize respiration flow rates. This step was achieved by coaching the subject to perform three tidal breaths through the device from which a within-session relative FRC reference was estimated from the end-expiratory level of the tidal breathing waveform. After this step, respiration flows and volumes could be continuously measured through coaching, delivery, and scan.

To investigate how coaching practices affect lung inflation and determine typical deviations from their relative FRC at ¹²⁹Xe inhalation, participants in Cohort A were coached to twice take a breath in and let it out, after which the coordinator actuated the device piston to initiate dose delivery. They then directed the participant to inhale the contents of the bag and hold their breath for the duration of the scan (about 8–10 seconds). Since the primary objective of this study was to observe subject behavior during the procedure, these measurements were performed using bags of air with the participants positioned on the patient table outside the MRI bore. However, five subjects did perform additional sessions with ¹²⁹Xe imaging later in the study for preliminary analysis on the effects of lung inflation on ¹²⁹Xe MRI repeatability. Each subject performed two to three sessions of coached breathing maneuvers followed by inhalation of the bag contents while gas flows were monitored and recorded. Additionally, subjects underwent pulmonary function testing (PFT), including spirometry and diffusing capacity of the lung for carbon monoxide (DLCO), as part of typical ¹²⁹Xe MRI procedure. To facilitate comparison between subjects, respiratory volumes were normalized to predicted vital capacity (VC), calculated as GLI-estimated total lung capacity (TLC) minus GLI-estimated residual volume (RV). Delivery error was defined as the difference between lung volume immediately before dose inhalation and the within-session relative FRC reference. For descriptive classification, dose inhalations beginning within ±5% of the subject’s GLI-estimated FRC were considered near target.

Measuring Alveolar Pressure Effects during ¹²⁹Xe MRI

To measure the effects of alveolar pressure conditions on the observed RBC:M, participants in Cohort B underwent five or six sequential spectroscopy scans using the pressure-sensing dose-delivery device. Dynamic spectroscopy of the dissolved-phase ¹²⁹Xe signal was acquired per consortium standards, using 500 free induction decays of 512 points with 20μs dwell times, TR/TE = 15/0.46ms and 20° flip angle. This TR/flip combination produces the same steady-state magnetization as a 90° pulse applied at TR₉₀ = 250ms, ensuring the signal is confined to the alveolar gas exchange units [28]. These data were analyzed to quantify the RBC:M, the RBC chemical shift, and RBC oscillation amplitude [29]. ¹²⁹Xe was delivered using conventional breath coaching with personnel placing the device mouthpiece in the subject’s mouth after the final preparatory exhale.

Participants began each session with one or two normal breath-hold scans, then performed two scans with the Mueller maneuver and two with the Valsalva maneuver. The normal maneuver was conducted by the participant inhaling and breath-holding according to standard coaching methods and served as the baseline for how the subject would naturally hold their breath. The Mueller maneuver was achieved by continuing to inhale against the empty bag, inducing negative alveolar pressure. The Valsalva maneuver was achieved by subjects attempting to exhale back into the bag while flow was blocked by the check valve. Throughout each of the scans, mouth pressure was continuously recorded. These pressures were analyzed to estimate their average deviations from the atmospheric value during the breath-hold maneuvers. Pressure deviations during the Mueller and Valsalva maneuvers were tested for their impact on RBC:M, quantified as percent change from the baseline normal scan.

Results

Respiration Monitoring

All 12 subjects in Cohort A successfully completed the coached breathing and inhalation maneuvers. A volume waveform collected from a representative subject is shown in Figure 5. It starts with tidal breathing, before progressing to the coached breaths, and culminating in the dose inhalation and the subsequent breath-hold; volumes are plotted relative to the mid-tidal volume determined during calibration, while delivery error was calculated relative to the within-session relative FRC reference. Most participants followed a similar pattern of progression from tidal to coached breathing with volume excursions increasing during the coached vs tidal breathing phase. Due to this behavior, the final coached exhalation dipped the subjects’ lung volumes below their relative FRC immediately prior to inhaling from the dose bag. This would ultimately result in a ¹²⁹Xe MRI acquisition at a volume below the target lung inflation.

Figure 5.

Representative volume waveforms measured during tidal breathing, preparatory breath coaching, xenon dose inhalation, and breath-hold. The volume of air respired over time is displayed relative to the tidal waveform midline measured during device calibration; the relative FRC reference was separately estimated from the end-expiratory level of the tidal breathing portion of the curve. In this example, we see that as the participant transitions from tidal breathing to coached breathing their inspiration and expiration volumes increase. The volume during exhalation tends to drop well below their relative functional residual capacity (FRC) as estimated by the minimum of the tidal volume portion of the curve. Ultimately, the participant begins their dose inhalation from roughly 467mL below their relative FRC, thereby causing them to fall well below the target lung volume during image acquisition.

Figure 6 summarizes breathing behaviors and lung volumes leading up to ¹²⁹Xe dose delivery. To enable comparison across subjects with different lung capacities, all volumes have been scaled to their VC. During tidal breathing, the average inhaled volume was 6.0%±3.7% of the subject’s VC, which was increased significantly during pre-dose coaching to 16.4%±4.7% of VC (p < 0.001). Consequently, lung volume at dose delivery was significantly biased below the target, with a mean delivery error of −7.7%±12.1% of VC (p < 0.001 vs zero error), indicating both systematic under-inflation relative to the target and substantial variability across deliveries. For an average VC of 4.95L±0.87L for the participants in this study, this would correspond to a mean delivery error of −381mL±599mL below their target lung inflation.

Figure 6.

A. Average breath volume of each individual dose delivery (N = 33) expressed as a percentage of the subject’s vital capacity (VC) during the tidal and coaching breathing phases prior to dose delivery. B. Lung volume error at dose delivery relative to the intended target volume (relative FRC), expressed as a percentage of VC. Most dose inhalations began at lung volumes below the target with substantial intersubject variability.

In this cohort, 18% of doses were delivered to subjects at their target volume within ±5% of their GLI-estimated FRC. Only 12% of doses were delivered above it with an average deviation of 447mL±146mL. The remaining 70% were delivered below the target with an average deviation of −564mL±411mL. As a result of these deviations, most subjects did not achieve their target lung inflation at the time of scanning.

To illustrate the effects of lung inflation, Figure 7 shows gas exchange MRI in a healthy volunteer scanned 800mL below the target lung volume of FRC+V_dose compared to one scanned at the target volume. The color-binned maps and quantitative gas-exchange metrics are presented based on thresholds derived from recently published healthy reference distributions [30]. The subject imaged below the target lung volume demonstrated elevated membrane signal intensity, a feature associated with interstitial lung disease (ILD). However, these findings were combined with the observation of high RBC transfer, which is not commonly observed in ILD. In contrast, the participant imaged at the target lung volume exhibited membrane and RBC transfer features expected for healthy volunteers.

Figure 7.

¹²⁹Xe MRI gas exchange analysis of two subjects with differing levels of lung inflation. Subject A was imaged at a lung volume 800mL below their target inflation of 3.00L; Subject B achieved a lung volume near FRC at delivery, resulting in a lung inflation only 30mL below the target of 2.83L. For each subject, representative slices of ventilation, membrane, and red blood cell (RBC) images are shown with corresponding histograms in the left column. Histogram y-axes indicate the fraction of total lung voxels. Histogram x-axes indicate signal intensity for each figure: ventilation was rescaled by the 99^th intensity percentile to a range of 0 to 1, while membrane and RBC signals are scaled as ratios to gas signal multiplied by 100. Dashed curves and color-bin thresholds are based on healthy reference distributions from Leewiwatwong et al [30]. While both were healthy volunteers, Subject A’s reduced lung inflation at the time of ¹²⁹Xe MRI produced high membrane uptake and elevated RBC transfer. Subject B met the expected lung inflation target and produced gas exchange features expected for their healthy status.

Alveolar Pressure During Breath-Hold

In Cohort B, all fourteen subjects completed normal, Mueller, and Valsalva maneuvers during spectroscopy MRI acquisitions while their mouth pressure was continuously monitored. This confirmed that the Mueller maneuver generated sub-atmospheric alveolar pressures when compared to the normal breath-hold while the Valsalva maneuver increased alveolar pressure by up to 25mmHg. A representative waveform collected from a subject performing both Mueller and Valsalva maneuvers is shown in Figure 8.

Figure 8.

A representative plot of mouth pressure changes during practice sessions of the Mueller and Valsalva maneuvers. The subject inhaled the full contents of the air bag and performed the Mueller maneuver for as long as they could. They then immediately switched to the Valsalva maneuver. As shown, pressure during these maneuvers can reach a magnitude up to 25mmHg above or below atmospheric pressure.

Across the cohort, the three breath-hold maneuvers achieved distinct alveolar pressures, with mean ± standard deviation values of −10.4 ± 8.2 mmHg during the normal maneuver, −14.5 ± 6.1 mmHg during the Mueller maneuver (p = 0.319 vs. normal), and +14.4 ± 4.7 mmHg during the Valsalva maneuver (p < 0.001 vs. normal). These pressure differences produced corresponding inverse changes in RBC:M (Figure 9). Relative to the normal maneuver (RBC:M = 0.446 ± 0.089), the Mueller maneuver significantly increased RBC:M to 0.464 ± 0.084 (p = 0.009), whereas the Valsalva maneuver decreased RBC:M to 0.409 ± 0.072 (p = 0.085). On average, Mueller increased RBC:M by 4.4%, while Valsalva decreased it by 7.0%. Both maneuvers also reduced RBC oscillation amplitudes relative to baseline, by 21.2% during Mueller and 25.8% during Valsalva.

Figure 9.

Average changes in estimated alveolar pressure, RBC:Membrane ratio (RBC:M), and percent change in RBC:M across breath-hold maneuvers. Pressure changes are quantified as mmHg relative to atmospheric pressure. Data are shown as mean values; error bars for ΔP indicate the standard deviation across subjects for each maneuver. The normal breath-hold served as each subject’s baseline for RBC:M and quantified the pressure deviation generated during standard dose inhalation and breath-hold. The Mueller maneuver, while associated with only a small reduction in alveolar pressure from normal (p = 0.319), exhibited a significant increase in RBC:M (p = 0.009). By contrast, the Valsalva maneuver significantly increased alveolar pressure from normal (p < 0.001), while RBC:M decreased but was not statistically significant (p = 0.085).

Figure 10 illustrates the relationship between estimated alveolar pressure and RBC:M across maneuvers. In Figure 10A, ΔPressure is expressed relative to atmospheric pressure for all breath-holds, with repeated measurements connected within each subject. This representation demonstrates the overall inverse pressure-response pattern: sub-atmospheric pressures generated during Mueller maneuvers were generally associated with increased RBC:M, whereas supra-atmospheric pressures generated during Valsalva maneuvers were generally associated with reduced RBC:M relative to normal breath-hold. Figure 10B summarizes these intra-subject changes by plotting ΔRBC:M relative to each subject’s normal maneuver. Variability in ΔRBC:M was greater during Valsalva than Mueller (SD 0.07 vs 0.02), with a larger mean decrease during Valsalva (mean ΔRBC:M ≈ −0.047) compared with the mean increase during Mueller (mean ΔRBC:M ≈ +0.014). Figure 10C further quantifies the pressure-response relationship using intra-subject slopes, Δ(RBC:M)/ΔP, calculated relative to each subject’s normal maneuver. These slopes were predominantly negative across subjects, indicating that increases in estimated alveolar pressure tended to reduce RBC:M, with steeper pressure-response slopes observed during the Mueller maneuver.

Figure 10.

A. RBC:M and alveolar pressure changes from atmospheric pressure for all dose deliveries across subjects. Each color represents an individual subject; symbols denote breathing maneuver and lines connect measurements within the same subject to emphasize directionality of change. Negative alveolar pressure changes during the Mueller maneuver are associated with increased RBC:M, whereas positive pressures during Valsalva are associated with reduced RBC:M relative to normal breathing. B. Violin plots show the distribution of intra-subject ΔRBC:M relative to each subject’s normal maneuver for Mueller and Valsalva breath-holds. Points indicate individual breath-holds and filled black circles denote group means. C. Violin plots show intra-subject pressure-response slopes (Δ(RBC:M)/ ΔP) for Mueller and Valsalva conditions. These were calculated with paired RBC:M and alveolar pressure data relative to the subject’s normal maneuver. Points indicate individual slopes and filled black circles denote group means.

Discussion

We demonstrated a modular ¹²⁹Xe dose delivery and physiological monitoring system with two complementary device configurations. The remotely actuated, Venturi-equipped configuration enabled room air breathing before dose release, dose delivery from outside the scanner bore, and continuous monitoring of breathing patterns throughout the procedure. These results establish the general utility of the system and support its potential translation as an assistive tool to guide subjects in real time toward target lung inflation for ¹²⁹Xe MRI. The pressure-sensing mouthpiece configuration, used with conventional personnel-assisted delivery, enabled mouth pressure monitoring and alveolar pressure estimation during the acquisition breath-hold. Together, these studies show that both lung volume at the time of dose inhalation and alveolar pressure during the subsequent breath-hold introduce substantial variability into ¹²⁹Xe MRI/MRS, highlighting patient performance and coaching consistency as key determinants of metric repeatability.

Variability in Lung Inflation and Its Consequences

We observed that standard coaching practices led participants to substantially alter their breathing patterns during the preparatory breaths preceding ¹²⁹Xe dose inhalation. Most subjects exhaled below FRC during this phase, resulting in lung volumes during imaging that were well below the expected target inflation of FRC+V_dose and, in some cases, below FRC entirely. Such low lung inflation levels markedly increased the membrane uptake and RBC transfer signals, complicating interpretation of gas exchange metrics. This effect is of particular concern as elevated membrane uptake is increasingly regarded as a feature associated with ILD [21]. While perhaps in healthy volunteers such findings would be offset by the concomitant observation of elevated RBC transfer, it is an undesirable complication to image interpretation. These findings reinforce earlier observations by Garrison et al. that lung inflation during ¹²⁹Xe MRI strongly influences Membrane:Gas and RBC:Gas ratios [19], but extend those results to more typical imaging conditions where volume changes arise from standard coaching.

This result has two key implications. First, tailoring the dosing volume to a fraction of FVC or TLC, while important, does not by itself guarantee that the target lung inflation will be achieved during the scan. Although these breath volumes were expressed relative to upright VC estimations, and thus likely underestimate the fraction of available lung capacity used during supine imaging [31], the participant’s lung volume at the onset of dose inhalation appears to be a larger source of variability. Second, the variability introduced is not random, but systematic: current coaching protocols actively drive subjects below their FRC prior to inhaling the dose. Although the current protocol explicitly instructs personnel to coach subjects through normal breathing using visual assessment of chest wall motion [18], this cue is not readily distinguishable during tidal breathing in many subjects with lower vital capacities. In addition, the act of consciously controlling breathing during coaching may itself perturb the subject’s natural resting breathing pattern, leading to systematic deviations from FRC at the time of dose inhalation. Together, these observations suggest that both subjective visual cueing and volitional breathing control can contribute to inconsistent lung inflation across subjects, highlighting an opportunity to further improve repeatability in ¹²⁹Xe MRI by incorporating more explicit and tangible guidance of lung inflation at the time of dose delivery.

Impact of Alveolar Pressure on Gas Exchange Metrics

This work also demonstrates that alveolar pressure during the acquisition breath-hold affects the measured RBC:M. Negative alveolar pressure induced by the Mueller maneuver increased RBC:M by over 4%, whereas positive pressure induced by the Valsalva maneuver decreased it by 7%. The magnitude of these changes is consistent with the reported repeatability of this metric in the literature [22,32].

These findings align with the role of alveolar pressure in modulating the transmural pressure gradient across capillary walls, thereby altering blood volume within the capillary bed [26]. Reduced alveolar pressures promote capillary recruitment and distension, enhancing ¹²⁹Xe uptake by RBCs. Conversely, elevated pressures compress the capillaries, reducing blood volume and thus diminishing RBC transfer.

Although the mean suppression of RBC:M during Valsalva was larger in magnitude, responses were substantially more variable across subjects (SD = 0.07). In contrast, Mueller-induced increases were more consistent, producing steeper and more uniform pressure-response relationships despite a smaller mean change. This greater heterogeneity during Valsalva likely reflects effort- and time-dependent alterations in venous return and pulmonary blood volume [33], which may amplify inter-subject variability under positive-pressure conditions. Thus, it is conceivable that inadvertent pressure maneuvers (i.e. Mueller- or Valsalva-like) during ¹²⁹Xe MRI/MRS could account for much of the RBC:M variability described in the published literature.

Beyond the effect on RBC:M, alveolar pressure had a surprisingly large effect on RBC oscillation amplitude, which was attenuated by ~20% of its original amplitude during both Mueller and Valsalva maneuvers. The attenuation at low alveolar pressure can be explained by increased capillary blood volume, which in turn experiences a smaller fractional change between systole and diastole [26]. The observation that a Valsalva maneuver also decreases oscillations is less intuitive, since it is expected to decrease the capillary blood pool and thus accentuate oscillatory phenomena. One interpretation is that the increasing alveolar pressure overwhelms the compliance-associated increase in capillary blood volume during systole. This may explain why RBC oscillation metrics have shown only modest repeatability with coefficients of variation around ±20% even under controlled conditions [22].

Conclusions

Integrating Lung Volume and Pressure Effects

Taken together, these studies demonstrate that lung inflation and alveolar pressure are two distinct but complementary sources of variability in ¹²⁹Xe MRI/MRS. Errors in lung inflation primarily alter the baseline partitioning of ¹²⁹Xe between gas exchange compartments, producing apparent differences in membrane uptake that can mimic lung disease. Pressure-related changes during the acquisition breath-hold dynamically affect pulmonary capillary blood volume and RBC oscillatory signals. These results highlight that unmonitored breathing behaviors during ¹²⁹Xe MRI can meaningfully influence measured metrics and potentially distort interpretation of clinically relevant findings.

Implications for Standardization and Clinical Translation

These findings illustrate the importance of continued efforts to incorporate physiological monitoring into dose delivery systems for ¹²⁹Xe MRI. The system described here provides three critical capabilities: remote actuation of dose delivery outside the scanner bore, real-time monitoring of inhalation volume to ensure dosing at or near FRC, and continuous pressure sensing during the breath-hold to identify abnormal maneuvers. This monitoring framework could form the basis for standardized operating procedures across sites, reducing intersession and inter-site variability. In future work, this approach could be expanded by integrating control systems that provide operator feedback when subjects deviate from expected tidal breathing or generate abnormal alveolar pressures, and by incorporating automated actuation to trigger dose delivery at predefined lung volume targets.

In a broader sense, this work highlights that physiological standardization is as important as technical standardization for quantitative pulmonary imaging. Just as harmonized acquisition protocols have been critical for multicenter reproducibility [18], harmonized coaching and physiological monitoring may be required to fully realize the biomarker potential of ¹²⁹Xe MRI.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request and with appropriate institutional review board approvals.