Source localization for neutron imaging systems using convolutional neural networks.

The nuclear imaging system at the National Ignition Facility (NIF) is a crucial diagnostic for determining the geometry of inertial confinement fusion implosions. The geometry is reconstructed from a neutron aperture image via a set of reconstruction algorithms using an iterative Bayesian inference approach. An important step in these reconstruction algorithms is finding the fusion source location within the camera field-of-view. Currently, source localization is achieved via an iterative optimization algorithm. In this paper, we introduce a machine learning approach for source localization. Specifically, we train a convolutional neural network to predict source locations given a neutron aperture image. We show that this approach decreases computation time by several orders of magnitude compared to the current optimization-based source localization while achieving similar accuracy on both synthetic data and a collection of recent NIF deuterium-tritium shots.

A TOPAS model for lens-based proton radiography.

Objective.Proton Radiography can be used in conjunction with proton therapy for patient positioning, real-time estimates of stopping power, and adaptive therapy in regions with motion. The modeling capability shown here can be used to evaluate lens-based radiography as an instantaneous proton-based radiographic technique. The utilization of user-friendly Monte Carlo program TOPAS enables collaborators and other users to easily conduct medical- and therapy- based simulations of the Los Alamos Neutron Science Center (LANSCE). The resulting transport model is an open-source Monte Carlo package for simulations of proton and heavy ion therapy treatments and concurrent particle imaging.Approach.The four-quadrupole, magnetic lens system of the 800-MeV proton beamline at LANSCE is modeled in TOPAS. Several imaging and contrast objects were modelled to assess transmission at energies from 230-930 MeV and different levels of particle collimation. At different proton energies, the strength of the magnetic field was scaled according toßγ,the inverse product of particle relativistic velocity and particle momentum.Main results.Materials with high atomic number, Z, (gold, gallium, bone-equivalent) generated more contrast than materials with low-Z (water, lung-equivalent, adipose-equivalent). A 5-mrad collimator was beneficial for tissue-to-contrast agent contrast, while a 10-mrad collimator was best to distinguish between different high-Z materials. Assessment with a step-wedge phantom showed water-equivalent path length did not scale directly according to predicted values but could be mapped more accurately with calibration. Poor image quality was observed at low energies (230 MeV), but improved as proton energy increased, with sub-mm resolution at 630 MeV.Significance.Proton radiography becomes viable for shallow bone structures at 330 MeV, and for deeper structures at 630 MeV. Visibility improves with use of high-Z contrast agents. This modality may be particularly viable at carbon therapy centers with accelerators capable of delivering high energy protons and could be performed with carbon therapy.

Bootstrap estimation of the effect of instrument response function uncertainty on the reconstruction of fusion neutron sources.

Neutron imagers are important diagnostics for the inertial confinement fusion implosions at the National Ignition Facility. They provide two- and three-dimensional reconstructions of the neutron source shape that are key indicators of the overall performance. To interpret the shape results properly, it is critical to estimate the uncertainty in those reconstructions. There are two main sources of uncertainties: limited neutron statistics, leading to random errors in the reconstructed images, and incomplete knowledge of the instrument response function (the pinhole-dependent point spread function). While the statistical errors dominate the uncertainty for lower yield deuterium-tritium (DT) shots, errors due to the instrument response function dominate the uncertainty for DT yields on the order of 1016 neutrons or higher. In this work, a bootstrapping method estimates the uncertainty in a reconstructed image due to the incomplete knowledge of the instrument response function. The main reconstruction is created from the fixed collection of pinhole images that are best aligned with the neutron source. Additional reconstructions are then built using subsets of that collection of images. Variations in the shapes of these additional reconstructions originate solely from uncertainties in the instrument response function, allowing us to use them to provide an additional systematic uncertainty estimate.

Contrast-enhanced proton radiographic sensitivity limits for tumor detection.

Purpose: Proton radiography may guide proton therapy cancer treatments with beam's-eye-view anatomical images and a proton-based estimation of proton stopping power. However, without contrast enhancement, proton radiography will not be able to distinguish tumor from tissue. To provide this contrast, functionalized, high- Z nanoparticles that specifically target a tumor could be injected into a patient before imaging. We conducted this study to understand the ability of gold, as a high- Z , biologically compatible tracer, to differentiate tumors from surrounding tissue. Approach: Acrylic and gold phantoms simulate a tumor tagged with gold nanoparticles (AuNPs). Calculations correlate a given thickness of gold to levels of tumor AuNP uptake reported in the literature. An identity, × 3 , and × 7 proton magnifying lens acquired lens-refocused proton radiographs at the 800-MeV LANSCE proton beam. The effects of gold in the phantoms, in terms of percent density change, were observed as changes in measured transmission. Variable areal densities of acrylic modeled the thickness of the human body. Results: A 1 - µ m -thick gold strip was discernible within 1 cm of acrylic, an areal density change of 0.2%. Behind 20 cm of acrylic, a 40 - µ m gold strip was visible. A 1-cm-diameter tumor tagged with 1 × 10 5 50-nm AuNPs per cell has an amount of contrast agent embedded within it that is equivalent to a 65 - µ m thickness of gold, an areal density change of 0.63% in a tissue thickness of 20 cm, which is expected to be visible in a typical proton radiograph. Conclusions: We indicate that AuNP-enhanced proton radiography might be a feasible technology to provide image-guidance to proton therapy, potentially reducing off-target effects and sparing nearby tissue. These data can be used to develop treatment plans and clinical applications can be derived from the simulations.

Preclinical MRI to Quantify Pulmonary Disease Severity and Trajectories in Poorly Characterized Mouse Models: A Pedagogical Example Using Data from Novel Transgenic Models of Lung Fibrosis.

Structural remodeling in lung disease is progressive and heterogeneous, making temporally and spatially explicit information necessary to understand disease initiation and progression. While mouse models are essential to elucidate mechanistic pathways underlying disease, the experimental tools commonly available to quantify lung disease burden are typically invasive (e.g., histology). This necessitates large cross-sectional studies with terminal endpoints, which increases experimental complexity and expense. Alternatively, magnetic resonance imaging (MRI) provides information noninvasively, thus permitting robust, repeated-measures statistics. Although lung MRI is challenging due to low tissue density and rapid apparent transverse relaxation (T2* <1 ms), various imaging methods have been proposed to quantify disease burden. However, there are no widely accepted strategies for preclinical lung MRI. As such, it can be difficult for researchers who lack lung imaging expertise to design experimental protocols-particularly for novel mouse models. Here, we build upon prior work from several research groups to describe a widely applicable acquisition and analysis pipeline that can be implemented without prior preclinical pulmonary MRI experience. Our approach utilizes 3D radial ultrashort echo time (UTE) MRI with retrospective gating and lung segmentation is facilitated with a deep-learning algorithm. This pipeline was deployed to assess disease dynamics over 255 days in novel, transgenic mouse models of lung fibrosis based on disease-associated, loss-of-function mutations in Surfactant Protein-C. Previously identified imaging biomarkers (tidal volume, signal coefficient of variation, etc.) were calculated semi-automatically from these data, with an objectively-defined high signal volume identified as the most robust metric. Beyond quantifying disease dynamics, we discuss common pitfalls encountered in preclinical lung MRI and present systematic approaches to identify and mitigate these challenges. While the experimental results and specific pedagogical examples are confined to lung fibrosis, the tools and approaches presented should be broadly useful to quantify structural lung disease in a wide range of mouse models.

Validating in vivo hyperpolarized 129 Xe diffusion MRI and diffusion morphometry in the mouse lung.

PURPOSE: Diffusion and lung morphometry imaging using hyperpolarized gases are promising tools to quantify pulmonary microstructure noninvasively in humans and in animal models. These techniques assume the motion encoded is exclusively diffusive gas displacement, but the impact of cardiac motion on measurements has never been explored. Furthermore, although diffusion morphometry has been validated against histology in humans and mice using 3 He, it has never been validated in mice for 129 Xe. Here, we examine the effect of cardiac motion on diffusion imaging and validate 129 Xe diffusion morphometry in mice. THEORY AND METHODS: Mice were imaged using gradient-echo-based diffusion imaging, and apparent diffusion-coefficient (ADC) maps were generated with and without cardiac gating. Diffusion-weighted images were fit to a previously developed theoretical model using Bayesian probability theory, producing morphometric parameters that were compared with conventional histology. RESULTS: Cardiac gating had no significant impact on ADC measurements (dual-gating: ADC = 0.020 cm2 /s, single-gating: ADC = 0.020 cm2 /s; P = .38). Diffusion-morphometry-generated maps of ADC (mean, 0.0165 ± 0.0001 cm2 /s) and acinar dimensions (alveolar sleeve depth [h] = 44 µm, acinar duct radii [R] = 99 µm, mean linear intercept [Lm ] = 74 µm) that agreed well with conventional histology (h = 45 µm, R = 108 µm, Lm = 63 µm). CONCLUSION: Cardiac motion has negligible impact on 129 Xe ADC measurements in mice, arguing its impact will be similarly minimal in humans, where relative cardiac motion is reduced. Hyperpolarized 129 Xe diffusion morphometry accurately and noninvasively maps the dimensions of lung microstructure, suggesting it can quantify the pulmonary microstructure in mouse models of lung disease.

Mapping and correcting hyperpolarized magnetization decay with radial keyhole imaging.

PURPOSE: Hyperpolarized (HP) media enable biomedical imaging applications that cannot be achieved with conventional MRI contrast agents. Unfortunately, quantifying HP images is challenging, because relaxation and radio-frequency pulsing generate spatially varying signal decay during acquisition. We demonstrate that, by combining center-out k-space sampling with postacquisition keyhole reconstruction, voxel-by-voxel maps of regional HP magnetization decay can be generated with no additional data collection. THEORY AND METHODS: Digital phantom, HP 129 Xe phantom, and in vivo 129 Xe human (N = 4 healthy; N = 2 with cystic fibrosis) imaging was performed using radial sampling. Datasets were reconstructed using a postacquisition keyhole approach in which 2 temporally resolved images were created and used to generate maps of regional magnetization decay following a simple analytical model. RESULTS: Mean, keyhole-derived decay terms showed excellent agreement with the decay used in simulations (R2 = 0.996) and with global attenuation terms in HP 129 Xe phantom imaging (R2 > 0.97). Mean regional decay from in vivo imaging agreed well with global decay values and displayed spatial heterogeneity that matched expected variations in flip angle and oxygen partial pressure. Moreover, these maps could be used to correct variable signal decay across the image volume. CONCLUSIONS: We have demonstrated that center-out trajectories combined with keyhole reconstruction can be used to map regional HP signal decay and to quantitatively correct images. This approach may be used to improve the accuracy of quantitative measures obtained from hyperpolarized media. Although validated with gaseous HP 129 Xe in this work, this technique can be generalized to any hyperpolarized agent.

A liquid VI scintillator cell for fast-gated neutron imaging.

The design of a new fast-gated neutron imaging system for the National Ignition Facility with much stricter timing constraints than a previous system has prompted the search for a fast scintillator material that can be used in imaging. A novel imaging cell based on Liquid VI has recently been developed with Eljen Technology and characterized at the Special Technologies Laboratory and the Los Alamos Neutron Science Center. The results show superior timing characteristics and spatial resolution, and sufficient light production for the new system compared to fast plastic scintillators previously used in neutron imaging. While the primary application is in inertial confinement fusion diagnostics, the imaging cell can be used in any fast-gated imaging application where timing characteristics and spatial resolution are of concern.

Inverse-collimated proton radiography for imaging thin materials.

Relativistic, magnetically focused proton radiography was invented at Los Alamos National Laboratory using the 800 MeV LANSCE beam and is inherently well-suited to imaging dense objects, at areal densities >20 g cm-2. However, if the unscattered portion of the transmitted beam is removed at the Fourier plane through inverse-collimation, this system becomes highly sensitive to very thin media, of areal densities <100 mg cm-2. Here, this inverse-collimation scheme is described in detail and demonstrated by imaging Xe gas with a shockwave generated by an aluminum plate compressing the gas at Mach 8.8. With a 5-mrad inverse collimator, an areal density change of just 49 mg cm-2 across the shock front is discernible with a contrast-to-noise ratio of 3. Geant4 modeling of idealized and realistic proton transports can guide the design of inverse-collimators optimized for specific experimental conditions and show that this technique performs better for thin targets with reduced incident proton beam emittance. This work increases the range of areal densities to which the system is sensitive to span from ∼25 mg cm-2 to 100 g cm-2, exceeding three orders of magnitude. This enables the simultaneous imaging of a dense system as well as thin jets and ejecta material that are otherwise difficult to characterize with high-energy proton radiography.

Single-breath clinical imaging of hyperpolarized (129)Xe in the airspaces, barrier, and red blood cells using an interleaved 3D radial 1-point Dixon acquisition.

PURPOSE: We sought to develop and test a clinically feasible 1-point Dixon, three-dimensional (3D) radial acquisition strategy to create isotropic 3D MR images of (129)Xe in the airspaces, barrier, and red blood cells (RBCs) in a single breath. The approach was evaluated in healthy volunteers and subjects with idiopathic pulmonary fibrosis (IPF). METHODS: A calibration scan determined the echo time at which (129)Xe in RBCs and barrier were 90° out of phase. At this TE, interleaved dissolved and gas-phase images were acquired using a 3D radial acquisition and were reconstructed separately using the NUFFT algorithm. The dissolved-phase image was phase-shifted to cast RBC and barrier signal into the real and imaginary channels such that the image-derived RBC:barrier ratio matched that from spectroscopy. The RBC and barrier images were further corrected for regional field inhomogeneity using a phase map created from the gas-phase (129)Xe image. RESULTS: Healthy volunteers exhibited largely uniform (129)Xe-barrier and (129)Xe-RBC images. By contrast, (129)Xe-RBC images in IPF subjects exhibited significant signal voids. These voids correlated qualitatively with regions of fibrosis visible on CT. CONCLUSIONS: This study illustrates the feasibility of acquiring single-breath, 3D isotropic images of (129)Xe in the airspaces, barrier, and RBCs using a 1-point Dixon 3D radial acquisition.

Dose and pulse sequence considerations for hyperpolarized (129)Xe ventilation MRI.

PURPOSE: The aim of this study was to evaluate the effect of hyperpolarized (129)Xe dose on image signal-to-noise ratio (SNR) and ventilation defect conspicuity on both multi-slice gradient echo and isotropic 3D-radially acquired ventilation MRI. MATERIALS AND METHODS: Ten non-smoking older subjects (ages 60.8±7.9years) underwent hyperpolarized (HP) (129)Xe ventilation MRI using both GRE and 3D-radial acquisitions, each tested using a 71ml (high) and 24ml (low) dose equivalent (DE) of fully polarized, fully enriched (129)Xe. For all images SNR and ventilation defect percentage (VDP) were calculated. RESULTS: Normalized SNR (SNRn), obtained by dividing SNR by voxel volume and dose was higher for high-DE GRE acquisitions (SNRn=1.9±0.8ml(-2)) than low-DE GRE scans (SNRn=0.8±0.2ml(-2)). Radially acquired images exhibited a more consistent, albeit lower SNRn (High-DE: SNRn=0.5±0.1ml(-2), low-DE: SNRn=0.5±0.2ml(-2)). VDP was indistinguishable across all scans. CONCLUSIONS: These results suggest that images acquired using the high-DE GRE sequence provided the highest SNRn, which was in agreement with previous reports in the literature. 3D-radial images had lower SNRn, but have advantages for visual display, monitoring magnetization dynamics, and visualizing physiological gradients. By evaluating normalized SNR in the context of dose-equivalent formalism, it should be possible to predict (129)Xe dose requirements and quantify the benefits of more efficient transmit/receive coils, field strengths, and pulse sequences.

Extending semiautomatic ventilation defect analysis for hyperpolarized (129)Xe ventilation MRI.

RATIONALE AND OBJECTIVES: Clinical deployment of hyperpolarized (129)Xe magnetic resonance imaging requires accurate quantification and visualization of the ventilation defect percentage (VDP). Here, we improve the robustness of our previous semiautomated analysis method to reduce operator dependence, correct for B1 inhomogeneity and vascular structures, and extend the analysis to display multiple intensity clusters. MATERIALS AND METHODS: Two segmentation methods were compared-a seeded region-growing method, previously validated by expert reader scoring, and a new linear-binning method that corrects the effects of bias field and vascular structures. The new method removes nearly all operator interventions by rescaling the (129)Xe magnetic resonance images to the 99th percentile of the cumulative distribution and applying fixed thresholds to classify (129)Xe voxels into four clusters: defect, low, medium, and high intensity. The methods were applied to 24 subjects including patients with chronic obstructive pulmonary disease (n = 8), age-matched controls (n = 8), and healthy normal subjects (n = 8). RESULTS: Linear-binning enabled a faster and more reproducible workflow and permitted analysis of an additional 0.25 ± 0.18 L of lung volume by accounting for vasculature. Like region-growing, linear-binning VDP correlated strongly with reader scoring (R(2) = 0.93, P < .0001), but with less systematic bias. Moreover, linear-binning maps clearly depict regions of low and high intensity that may prove useful for phenotyping subjects with chronic obstructive pulmonary disease. CONCLUSIONS: Corrected linear-binning provides a robust means to quantify (129)Xe ventilation images yielding VDP values that are indistinguishable from expert reader scores, while exploiting the entire dynamic range to depict multiple image clusters.

Measuring diffusion limitation with a perfusion-limited gas--hyperpolarized 129Xe gas-transfer spectroscopy in patients with idiopathic pulmonary fibrosis.

Although xenon is classically taught to be a "perfusion-limited" gas, (129)Xe in its hyperpolarized (HP) form, when detected by magnetic resonance (MR), can probe diffusion limitation. Inhaled HP (129)Xe diffuses across the pulmonary blood-gas barrier, and, depending on its tissue environment, shifts its resonant frequency relative to the gas-phase reference (0 ppm) by 198 ppm in tissue/plasma barrier and 217 ppm in red blood cells (RBCs). In this work, we hypothesized that in patients with idiopathic pulmonary fibrosis (IPF), the ratio of (129)Xe spectroscopic signal in the RBCs vs. barrier would diminish as diffusion-limitation delayed replenishment of (129)Xe magnetization in RBCs. To test this hypothesis, (129)Xe spectra were acquired in 6 IPF subjects as well as 11 healthy volunteers to establish a normal range. The RBC:barrier ratio was 0.55 ± 0.13 in healthy volunteers but was 3.3-fold lower in IPF subjects (0.16 ± 0.03, P = 0.0002). This was caused by a 52% reduction in the RBC signal (P = 0.02) and a 58% increase in the barrier signal (P = 0.01). Furthermore, the RBC:barrier ratio strongly correlated with lung diffusing capacity for carbon monoxide (DLCO) (r = 0.89, P < 0.0001). It exhibited a moderate interscan variability (8.25%), and in healthy volunteers it decreased with greater lung inflation (r = -0.78, P = 0.005). This spectroscopic technique provides a noninvasive, global probe of diffusion limitation and gas-transfer impairment and forms the basis for developing 3D MR imaging of gas exchange.

Enabling hyperpolarized (129) Xe MR spectroscopy and imaging of pulmonary gas transfer to the red blood cells in transgenic mice expressing human hemoglobin.

PURPOSE: Hyperpolarized (HP) (129) Xe gas in the alveoli can be detected separately from (129) Xe dissolved in pulmonary barrier tissues (blood plasma and parenchyma) and red blood cells (RBCs) of humans, allowing this isotope to probe impaired gas uptake. Unfortunately, mice, which are favored as lung disease models, do not display a unique RBC resonance, thus limiting the preclinical utility of (129) Xe MR. Here we overcome this limitation using a commercially available strain of transgenic mice that exclusively expresses human hemoglobin. METHODS: Dynamic HP (129) Xe MR spectroscopy, and three-dimensional radial MRI of gaseous and dissolved (129) Xe were performed in both wild-type (C57BL/6) and transgenic mice. RESULTS: Unlike wild-type animals, transgenic mice displayed two dissolved (129) Xe NMR peaks at 198 and 217 ppm, corresponding to (129) Xe dissolved in barrier tissues and RBCs, respectively. Moreover, signal from these resonances could be imaged separately, using a 1-point variant of the Dixon technique. CONCLUSION: It is now possible to examine the dynamics and spatial distribution of pulmonary gas uptake by the RBCs of mice using HP (129) Xe MR spectroscopy and imaging. When combined with ventilation imaging, this ability will enable translational "mouse-to-human" studies of impaired gas exchange in a variety of pulmonary diseases.

Probing the regional distribution of pulmonary gas exchange through single-breath gas- and dissolved-phase 129Xe MR imaging.

Although some central aspects of pulmonary function (ventilation and perfusion) are known to be heterogeneous, the distribution of diffusive gas exchange remains poorly characterized. A solution is offered by hyperpolarized 129Xe magnetic resonance (MR) imaging, because this gas can be separately detected in the lung's air spaces and dissolved in its tissues. Early dissolved-phase 129Xe images exhibited intensity gradients that favored the dependent lung. To quantitatively corroborate this finding, we developed an interleaved, three-dimensional radial sequence to image the gaseous and dissolved 129Xe distributions in the same breath. These images were normalized and divided to calculate "129Xe gas-transfer" maps. We hypothesized that, for healthy volunteers, 129Xe gas-transfer maps would retain the previously observed posture-dependent gradients. This was tested in nine subjects: when the subjects were supine, 129Xe gas transfer exhibited a posterior-anterior gradient of -2.00 ± 0.74%/cm; when the subjects were prone, the gradient reversed to 1.94 ± 1.14%/cm (P < 0.001). The 129Xe gas-transfer maps also exhibited significant heterogeneity, as measured by the coefficient of variation, that correlated with subject total lung capacity (r = 0.77, P = 0.015). Gas-transfer intensity varied nonmonotonically with slice position and increased in slices proximal to the main pulmonary arteries. Despite substantial heterogeneity, the mean gas transfer for all subjects was 1.00 ± 0.01 while supine and 1.01 ± 0.01 while prone (P = 0.25), indicating good "matching" between gas- and dissolved-phase distributions. This study demonstrates that single-breath gas- and dissolved-phase 129Xe MR imaging yields 129Xe gas-transfer maps that are sensitive to altered gas exchange caused by differences in lung inflation and posture.

Quantitative analysis of hyperpolarized 129Xe ventilation imaging in healthy volunteers and subjects with chronic obstructive pulmonary disease.

In this study, hyperpolarized (129) Xe MR ventilation and (1) H anatomical images were obtained from three subject groups: young healthy volunteers (HVs), subjects with chronic obstructive pulmonary disease (COPD) and age-matched controls (AMCs). Ventilation images were quantified by two methods: an expert reader-based ventilation defect score percentage (VDS%) and a semi-automated segmentation-based ventilation defect percentage (VDP). Reader-based values were assigned by two experienced radiologists and resolved by consensus. In the semi-automated analysis, (1) H anatomical images and (129) Xe ventilation images were both segmented following registration to obtain the thoracic cavity volume and ventilated volume, respectively, which were then expressed as a ratio to obtain the VDP. Ventilation images were also characterized by generating signal intensity histograms from voxels within the thoracic cavity volume, and heterogeneity was analyzed using the coefficient of variation (CV). The reader-based VDS% correlated strongly with the semi-automatically generated VDP (r = 0.97, p < 0.0001) and with CV (r = 0.82, p < 0.0001). Both (129) Xe ventilation defect scoring metrics readily separated the three groups from one another and correlated significantly with the forced expiratory volume in 1 s (FEV1 ) (VDS%: r = -0.78, p = 0.0002; VDP: r = -0.79, p = 0.0003; CV: r = -0.66, p = 0.0059) and other pulmonary function tests. In the healthy subject groups (HVs and AMCs), the prevalence of ventilation defects also increased with age (VDS%: r = 0.61, p = 0.0002; VDP: r = 0.63, p = 0.0002). Moreover, ventilation histograms and their associated CVs distinguished between subjects with COPD with similar ventilation defect scores, but visibly different ventilation patterns.

In vivo MR imaging of pulmonary perfusion and gas exchange in rats via continuous extracorporeal infusion of hyperpolarized 129Xe.

BACKGROUND: Hyperpolarized (HP) (129)Xe magnetic resonance imaging (MRI) permits high resolution, regional visualization of pulmonary ventilation. Additionally, its reasonably high solubility (>10%) and large chemical shift range (>200 ppm) in tissues allow HP (129)Xe to serve as a regional probe of pulmonary perfusion and gas transport, when introduced directly into the vasculature. In earlier work, vascular delivery was accomplished in rats by first dissolving HP (129)Xe in a biologically compatible carrier solution, injecting the solution into the vasculature, and then detecting HP (129)Xe as it emerged into the alveolar airspaces. Although easily implemented, this approach was constrained by the tolerable injection volume and the duration of the HP (129)Xe signal. METHODS AND PRINCIPAL FINDINGS: Here, we overcome the volume and temporal constraints imposed by injection, by using hydrophobic, microporous, gas-exchange membranes to directly and continuously infuse (129)Xe into the arterial blood of live rats with an extracorporeal (EC) circuit. The resulting gas-phase (129)Xe signal is sufficient to generate diffusive gas exchange- and pulmonary perfusion-dependent, 3D MR images with a nominal resolution of 2×2×2 mm(3). We also show that the (129)Xe signal dynamics during EC infusion are well described by an analytical model that incorporates both mass transport into the blood and longitudinal relaxation. CONCLUSIONS: Extracorporeal infusion of HP (129)Xe enables rapid, 3D MR imaging of rat lungs and, when combined with ventilation imaging, will permit spatially resolved studies of the ventilation-perfusion ratio in small animals. Moreover, EC infusion should allow (129)Xe to be delivered elsewhere in the body and make possible functional and molecular imaging approaches that are currently not feasible using inhaled HP (129)Xe.