Junlan Lu¹, Suphachart Leewiwatwong², David Mummy³, Elianna Bier², and Bastiaan Driehuys^3
1Medical Physics, Duke University, Durham, NC, United States, ²Biomedical Engineering, Duke University, Durham, NC, United States, ³Radiology, Duke University, Durham, NC, United States

Although hyperpolarized ¹²⁹Xe gas exchange MRI enables imaging ventilation, barrier, and RBC components in a single breath-hold, the necessary under-sampling imposed by limited imaging time constrains image resolution. Therefore, it is common to acquire an additional dedicated ventilation scan, which increases cost and imaging time. Instead, we demonstrate that deep convolutional neural networks with template-based augmentation can be trained to transform under-sampled low-resolution ¹²⁹Xe ventilation images to a level of detail comparable to that of a dedicated ventilation scan. We evaluate the performance of multiple super-resolution models based on signal-to-noise ratio and structural similarity.

Abdullah S. Bdaiwi^1,2, Peter J. Nedbalski¹, Md M. Hossain³, Matthew M. Willmering¹, Laura L. Walkup^1,2,4, Hui Wang⁵, Robert P. Thomen¹, Kai Ruppert¹, Jason C. Woods^1,4,6, and Zackary I. Cleveland^1,2,4,6
1Center for Pulmonary Imaging Research, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ²Biomedical Engineering Department, University of Cincinnati, Cincinnati, OH, United States, ³Division of Biostatistics and Epidemiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ⁴Division of Pulmonary Medicine, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, United States, ⁵Philips, Cincinnati, OH, United States, ⁶Department of Pediatrics, University of Cincinnati, Cincinnati, OH, United States

Hyperpolarized (HP) ¹²⁹Xe MRI can non-invasively quantify lung microstructure through measuring the gas-phase apparent diffusion coefficient (ADC), providing insights into disease pathophysiology—e.g., emphysema severity. To maximize the utility of this method, we developed an analytical model, based on HP-specific Bloch equations and error propagation, to optimize ADC measurements in terms of acquisition parameters (b-value, flip angle, phase encodes, etc.) and the ADC itself and validated this model via Monte Carlo and phantom studies. Finally, a lower bound on the expected ¹²⁹Xe ADC was obtained by measuring ADC as a function of age in human subjects as young as six years.

Mu He¹, Lindsay A. Somerville¹, Nicolas J. Tustison¹, James Patrie¹, Jaime F. Mata¹, Joanne M. Cassani², Roselove Nunoo-Asare¹, Alan Ropp¹, Wilson G. Miller¹, Yun Michael Shim¹, Talissa A. Altes², John P. Mulger¹, and Eduard E. de Lange^1
1University of Virginia, Charlottesville, VA, United States, ²University of Missouri, Columbia, MO, United States

With hyperpolarized helium-3 (³He) MRI regional differences in airflow heterogeneity can be assessed. In this study, we used weighted entropy to evaluate dynamic ventilation heterogeneity in asthmatics. Serial ³He/¹H scans were co-registered and normalized for comparison. Weighted entropy was calculated based on entropy and Ventilation defect percentage (VDP) values derived from normalized ³He scans. Lobar analysis was performed to identify lobar weighted entropy. We found that asthmatics have more heterogeneous ventilation distribution in the lower lobes at baseline. Variable airflow heterogeneity changes were observed globally in asthmatics, with the most significant bronchodilator response in the left lung.

Joshua R Astley^1,2, Alberto M Biancardi¹, Paul J Hughes¹, Laurie J Smith¹, Helen Marshall¹, Guilhem J Collier¹, James Eaden¹, Nicholas D Weatherley¹, Jim M Wild¹, and Bilal A Tahir^1,2
1POLARIS, University of Sheffield, Sheffield, United Kingdom, ²Oncology and Metabolism, University of Sheffield, Sheffield, United Kingdom

Deep learning has shown great promise for numerous medical image segmentation tasks, including delineation of ventilated lung volumes from hyperpolarized gas MRI. We previously demonstrated the utility of a VNet convolutional neural network (CNN), trained on a combination of ³He and ¹²⁹Xe scans, in producing accurate segmentations that outperform conventional methods. In this work, we compared the performance of several 3D CNNs and loss functions for segmentation of ventilated lungs on a significantly larger and more diverse multi-nuclear hyperpolarised gas MRI dataset using several training strategies. We observe that the UNet CNN provides the best performing model for our dataset.

Maksym Sharma¹, Alexander M Matheson¹, David G McCormack², David A Palma^1,3, and Grace Parraga^1,2,3
1Medical Biophysics, Western University, London, ON, Canada, ²Division of Respirology, Department of Medicine, Western University, London, ON, Canada, ³Department of Oncology, Western University, London, ON, Canada

Pulmonary hyperpolarized ³He MRI provides a way to measure lung ventilation heterogeneity in patients with COPD, including terminal airspace enlargement or emphysema that is typically quantified using CT densitometry. Unfortunately, MRI-derived biomarkers of emphysema progression remain unconfirmed, and also likely because of radiation dose considerations, CT follow-up of emphysema is rarely performed, and hence its longitudinal progression is not well-understood. Here we developed a machine-learning pipeline that identified hyperpolarized ³He MRI texture features that independently and uniquely correlated and predicted rapidly-worsening emphysema nearly 3 years later, measured as CT RA₉₅₀, using a Decision Tree algorithm that achieved 82% prediction accuracy.

Rolf F Schulte¹, Guilhem J Collier², James Ball², Graham Norquay², Madhwesha Rao², and Jim M Wild^2
1GE Healthcare, Munich, Germany, ²POLARIS, Department of Infection Immunity & Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom

3D density-weighted MRSI in combination with a frequency-tailored RF excitation pulse was designed, implemented and used to detect xenon gas in the lungs and xenon dissolved in lung tissue and blood. These images were used to calculate quantitative ratio maps of tissue-to-gas, blood-to-gas, and blood-to-tissue with good SNR.

Ronald Pratt¹, Randy Giaquinto¹, Wolfgang Loew¹, Christopher Ireland¹, Neil Stewart^2,3, Nara Higano², Jason Woods², and Charles Dumoulin^1
1Imaging Research Center/Radiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ²Center for Pulmonary Imaging Research/Radiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ³POLARIS, Imaging Sciences, Dept of Infection, Immunity & Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom

Production of hyperpolarized ¹²⁹Xe gas has opened new frontiers for imaging lung anatomy and function using MRI.  Our institution has the goal of using hyperpolarized ¹²⁹Xe gas for MR imaging of neonatal lungs at 1.5 T.  To realize this goal, it is essential that the RF coil used for imaging operate at both the ¹²⁹Xe and ¹H frequencies.   This report provides fabrication details of a novel ¹²⁹Xe/¹H switched frequency transmit/receive 16 rung high-pass birdcage RF coil.  Images collected with hyperpolarized ¹²⁹Xe gas and water in phantoms demonstrate that the coil provides excellent image quality at both frequencies.

 

Graham Norquay¹, Guilhem J Collier¹, and Jim M Wild^1
1Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom

Global ¹²⁹Xe MRS was acquired in the lungs synchronously with an ECG recording to enable association of dissolved-phase ¹²⁹Xe signal dynamics with the cardiac cycle. Time-domain Voigt fitting was used to calculate resonance parameters corresponding to ¹²⁹Xe in red blood cells and tissue/blood plasma in the lungs. Positive and negative signal changes of ¹²⁹Xe in the blood were found to be associated with ventricular systole and diastole in a healthy volunteer.  

Faraz Amzajerdian¹, Tahmina Achekzai¹, Luis Loza¹, Hooman Hamedani¹, Yi Xin¹, Harilla Profka¹, Ian Duncan¹, Stephen Kadlecek¹, Kai Ruppert¹, and Rahim Rizi^1
1University of Pennsylvania, Philadelphia, PA, United States

Xenon-polarization Transfer Contrast (XTC) imaging is a powerful technique for quantifying exchange rates between gas- and dissolved-phase xenon compartments and involves a series of radiofrequency (RF) saturation pulses applied to the targeted dissolved-phase resonance. Although increasing the number of pulses and their spacing generates greater contrast, it also increases acquisition time. In this work, in an effort to reduce acquisition time, particularly for free-breathing imaging protocols, we explored the use of continuous RF irradiation to depolarize dissolved-phase xenon, with the goal of producing results similar to those achieved with pulsed saturations, but in a significantly shorter time period.

Hooman Hamedani¹, Stephen Kadlecek¹, Faraz Amzajerdian¹, Ryan Baron¹, Kai Ruppert¹, Ian Duncan¹, Yi Xin¹, Luis Loza¹, Tahmina Achekzai¹, Maurizio Cereda¹, Kevin Ma¹, David DiBardino¹, and Rahim Rizi^1
1University of Pennsylvania, Philadelphia, PA, United States

We have previously shown the advantages of multi- over single-breath imaging of lung ventilation using HP gas MRI, and have recently introduced our approach for comprehensively assessing the lung function during a tidal breathing scheme to quantify ventilation and dissolved parameters of xenon distribution. Here, we presented the repeatability of the imaging markers in a healthy subject and a patient with severe chronic obstructive pulmonary disease (COPD) with persisting cough. We have shown that the technical back-to-back reproducibility of measuring lung function with our scheme is excellent in a healthy subject and satisfactory in a COPD subject with persisting cough.

Madhwesha R Rao¹, Guilhem J Collier¹, Graham Norquay¹, Rolf F Schulte², and Jim M Wild^1
1University of Sheffield, Sheffield, United Kingdom, ²GE Healthcare, Munich, Germany

3D isotropic images of hyperpolarized ¹²⁹Xe dissolved in the brain were acquired with 1 L of inhaled xenon gas dose using a density-weighted MR spectroscopic imaging method. Images were acquired from 2 healthy male volunteers. The data was acquired with a voxel size of (2cm)³ and reconstructed to a voxel size of  (0.625 cm)³ through zero filling. The isotropic voxel size and the high spectral resolution surpass previous reports using conventional 2D and echo planar spectroscopic imaging methods.

Ho-Fung Chan¹, Guilhem J Collier¹, Madhwesha Rao¹, and Jim M Wild^1
1Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom

Background field inhomogeneities due to air-tissue interface magnetic susceptibility differences in the lungs increases derived ADC measurements from ³He PGSE diffusion-weighted (DW) lung MRI at higher field strengths. To determine if the same effects are observed with ¹²⁹Xe DW-MRI, ¹²⁹Xe ADC and mean diffusive length scale (Lm_D) from healthy volunteers imaged at 1.5T and 3T were compared. A small bias towards increased ¹²⁹Xe ADC (6.3%) and Lm_D (2.2%) values at 3T was obtained, which is smaller than the reported bias of ³He ADC (15.5%) and ³He mean chord length (10.5%), and similar to the reported ¹²⁹Xe DW-MRI inter-scan repeatability differences.

Kai Ruppert¹, Faraz Amzajerdian¹, Yi Xin¹, Hooman Hamedani¹, Luis Loza¹, Tahmina S Achekzai¹, Ryan J Baron¹, Ian F Duncan¹, Harrilla Profka¹, Yiwen Qian¹, Stephen Kadlecek¹, Alessandra Fusco², Benjamin Sinder³, Patrick J Cahill³, Brian Snyder^3,4, Thomas P Schaer², and Rahim R Rizi^1
1University of Pennsylvania, Philadelphia, PA, United States, ²School of Veterinary Medicine, University of Pennsylvania, Kennett Square, PA, United States, ³Children's Hospital of Philadelphia, Philadelphia, PA, United States, ⁴Boston Children's Hospital, Boston, MA, United States

Thoracic insufficiency syndrome (TIS) progresses to the development of restrictive lung disease and is commonly treated through surgical intervention. In this work, we used a rib-tether rabbit model to investigate the sensitivity of dynamic 1D simultaneous dissolved- and gas-phase hyperpolarized xenon-129 MRI imaging to pulmonary abnormalities secondary to TIS. We found asymmetric lung ventilation patterns and increases in alveolar septal wall thickness in both lungs of a rib-tethered rabbit compared to an age-matched control animal. These findings could help identify the optimal timepoint at which to conduct chest expansion surgery so as to maximize the resulting improvements in lung maturation.

James Ball¹, Jim M. Wild¹, and Graham Norquay^1
1POLARIS, Department of Infection, Immunity and Cardiovascular Disease, The University of Sheffield, Sheffield, United Kingdom

Hyperpolarised xenon-129 (¹²⁹Xe) production via continuous-flow spin-exchange optical pumping often produces lower than predicted ¹²⁹Xe polarisation. Frequently, thermodynamics within polariser systems is not considered, as well as spatially variant changes in rubidium (Rb) source distributions and changes in ¹²⁹Xe relaxation over time. This work models realistic Rb and temperature distributions within a large-scale ¹²⁹Xe-Rb polariser cell. Results show that Rb density varies significantly depending upon incident photon flux and Rb source distribution in the cell. Modelling allows estimation of the Rb presaturator length required at different gas flow rates in order to reach optimal Rb density and high ¹²⁹Xe polarisation.

Junlan Lu¹, David Mummy², Suphachart Leewiwatwong³, Elianna Bier³, and Bastiaan Driehuys^2
1Medical Physics, Duke University, Durham, NC, United States, ²Radiology, Duke University, Durham, NC, United States, ³Biomedical Engineering, Duke University, Durham, NC, United States

Accurately correcting hyperpolarized ¹²⁹Xe ventilation MRI for coil-induced bias field remains the most significant obstacle to precise, repeatable quantitative image analysis. Estimates using B1 maps from RF-depletion in radially acquired images may provide an improvement on the standard N4ITK solution, which recent works suggests may perform an overly aggressive correction. Here, we develop a template-based approach to bias field correction using B1 maps derived from multiple subjects. This paradigm is then evaluated by applying it to a set of test images reflective of a range of disease types and levels of ventilation obstruction.

Tahmina Susan Achekzai¹, Luis Loza², Stephen Kadlecek², Kai Ruppert², Faraz Amzajerdian², and Rahim R. Rizi^1
1Radiology, University of Pennsylvania, Philadelphia, PA, United States, ²University of Pennsylvania, Philadelphia, PA, United States

XTC imaging takes advantage of the exchange between gas and dissolved phase Xe to provide information on lung function, as direct dissolved phase imaging is limited in signal. In this study, we demonstrate XTC imaging in a regular and transgenic mouse model. We aim to optimize depolarization effectiveness by varying delay time and number of saturation pulses applied in the imaging scheme and demonstrate the dependence of depolarization to these parameters. 

Alexander M Matheson¹, Rachel L Eddy¹, Jonathan L MacNeil², Marrissa J McIntosh¹, and Grace M Parraga^1,2
1Medical Biophysics, Robarts Research Institute, Western University, London, ON, Canada, ²School of Biomedical Engineering, Robarts Research Institute, Western University, London, ON, Canada

Co-registered hyperpolarized gas and proton MRI are required to calculate functional lung biomarkers using semi-automated pipelines. Automated registration between different spectral acquisitions is difficult due to differences in contrast and imaging features between different spectral images. Convolutional neural networks create abstract representations of images that may overcome these feature differences. We retrospectively pooled data sets previously registered using a semi-automated pipeline and applied random two-dimensional affine transformations and noise. Neural networks generated inverse transformation matrices from these data to correct the applied transformations. The trained network successfully corrected mis-alignment with an average error of less than one pixel. 

Yurii Shepelytskyi^1,2, Vira Grynko^1,2, Tao Li³, Ayman Hassan^4,5, Karl Granberg⁴, and Mitchell S Albert^2,3,5
1Chemistry and Materials Science Program, Lakehead University, Thunder Bay, ON, Canada, ²Thunder Bay Regional Health Research Institute, Thunder Bay, ON, Canada, ³Chemistry, Lakehead University, Thunder Bay, ON, Canada, ⁴Thunder Bay Regional Health Sciences Centre, Thunder Bay, ON, Canada, ⁵Northern Ontario School of Medicine, Thunder Bay, ON, Canada

Hyperpolarized (HP) xenon-129 (¹²⁹Xe) freely dissolves in pulmonary blood and travels to highly perfused organs. Dissolved phase HP ¹²⁹Xe imaging is commonly used for evaluating gas-blood exchange in lungs, imaging cerebral perfusion, detecting hemodynamic response, and kidney perfusion. However, the signal-to-noise ratio (SNR) of HP ¹²⁹Xe dissolved phase images varies between breath-holds, especially for brain imaging. In this work, we demonstrated a significant reduction in the variability of MRI image SNR by implementing an additional depolarization pulse prior to image acquisition.

Abdullah S. Bdaiwi^1,2, Matthew M. Willmering¹, Hui Wang³, and Zackary I. Cleveland^1,2,4,5
1Center for Pulmonary Imaging Research, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ²Biomedical Engineering Department, University of Cincinnati, Cincinnati, OH, United States, ³Philips, Cincinnati, OH, United States, ⁴Department of Pediatrics, University of Cincinnati, Cincinnati, OH, United States, ⁵Division of Pulmonary Medicine, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, United States

Diffusion-weighted hyperpolarized ¹²⁹Xe MRI is a validated measure of lung microstructure and can assess changes in alveolar dimensions. These images are commonly acquired via 2D gradient recalled echo (GRE), but ¹²⁹Xe GRE images suffer from coarse resolution in the slice dimension and breath-hold durations (≤16 s) that may be difficult for pediatric and severely ill subjects. To overcome these limitations, we implemented 2D- and 3D-spiral (FLORET, Fermat looped, orthogonally encoded trajectories) sequences for ¹²⁹Xe diffusion imaging. These sequences enable either rapid acquisition or high-resolution, isotropic lung coverage and display image quality and ADC accuracy comparable to that of conventional 2D-GRE.

Rachel L Eddy^1,2, Alexander M Matheson^3,4, Marrissa J McIntosh^3,4, and Grace Parraga^3,4
1St. Paul's Hospital, UBC Centre for Heart Lung Innovation, Vancouver, BC, Canada, ²Division of Respiratory Medicine, Department of Medicine, University of British Columbia, Vancouver, BC, Canada, ³Robarts Research Institute, London, ON, Canada, ⁴Department of Medical Biophysics, Western University, London, ON, Canada

Pulmonary ventilation has been shown to follow a fractal distribution using fluorescence imaging. ¹²⁹Xe MRI provides high spatial-temporal resolution images of pulmonary ventilation so here, we aimed to determine the fractal properties of ¹²⁹Xe MRI ventilation heterogeneity using the box-counting method.  In 25 patients with asthma, MRI ventilation heterogeneity followed a power law and mean fractal dimension for MRI signal ranged from 1.39-1.82.  Fractal analysis can provide a new tool to measure regional MRI ventilation heterogeneity and investigate pulmonary structure-function relationships in patients with lung disease.