Joshua R Astley^1,2, Alberto M Biancardi², Helen Marshall², Paul JC Hughes², Guilhem J Collier², Laurie J Smith², James Eaden², François-Xavier Blé³, Rod Hughes⁴, Jim M Wild^2,5, and Bilal A Tahir^1,2
1Oncology and Metabolism, University of Sheffield, Sheffield, United Kingdom, ²POLARIS, Sheffield, United Kingdom, ³Translational Science and Experimental Medicine, Research and Early Development, Respiratory & Immunology, BioPharmaceuticals R&D, AstraZeneca, Cambridge, United Kingdom, ⁴Clinical Development, Research and Early Development, Respiratory & Immunology, AstraZeneca, Cambridge, United Kingdom, ⁵Insigneo Institute for in silico medicine, University of Sheffield, Sheffield, United Kingdom

Functional lung imaging biomarkers, such as the ventilated defect percentage, are computed from segmentations derived from spatially co-registered functional hyperpolarized gas MRI and structural ¹H-MRI. Although hyperpolarized gas and ¹H-MRI can be acquired in the same or similar breaths, the acquired scans are frequently misaligned. Here, we propose a multi-channel deep learning-based approach to generate accurate lung cavity estimations (LCEs). Across all evaluation metrics, the multi-channel approach significantly outperformed single-channel approaches and generated plausible LCEs across a wide range of pulmonary pathologies. In addition, correlation and Bland-Altman analyses of lung volumes demonstrated strong correlation and minimal bias with expert LCEs.

Joshua R Astley^1,2, Alberto M Biancardi², Helen Marshall², Malina-Maria Tofan¹, Laurie J Smith², Paul JC Hughes², Guilhem J Collier², Matthew Q Hatton^1,3, François-Xavier Blé⁴, Rod Hughes⁵, Jim M Wild^2,6, and Bilal A Tahir^1,2
1Oncology and Metabolism, University of Sheffield, Sheffield, United Kingdom, ²POLARIS, Sheffield, United Kingdom, ³Clinical Oncology, Weston Park Cancer Centre, Sheffield, United Kingdom, ⁴Translational Science and Experimental Medicine, Research and Early Development, Respiratory & Immunology, BioPharmaceuticals R&D, AstraZeneca, Cambridge, United Kingdom, ⁵Clinical Development, Research and Early Development, Respiratory & Immunology, AstraZeneca, Cambridge, United Kingdom, ⁶Insigneo Institute for in silico medicine, University of Sheffield, Sheffield, United Kingdom

Hyperpolarized gas MRI can visualize and quantify regional lung ventilation with exquisite detail, but clinical uptake is limited to a few centres worldwide. Alternative, non-contrast techniques have been proposed to image ventilation, including ¹H-based surrogates of ventilation from multi-inflation ¹H-MRI. Recently, deep learning has shown potential for generating synthetic images in multiple modalities within the lung image analysis field. We propose a 3D multi-channel deep learning approach to synthesize hyperpolarized gas MRI and assess its quantitative performance using several common image synthesis metrics across a large, diverse dataset of lung pathologies using 5-fold cross-validation.

Ahsan Javed¹, Rajiv Ramasawmy¹, Joel Moss², Waqas Majeed³, Pan Su³, Thomas Benkert⁴, Himanshu Bhat³, Ashkan Malayeri⁵, and Adrienne E. Campbell-Washburn^1
1Cardiovascular Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, United States, ²Pulmonary Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, United States, ³Siemens Medical Solutions USA Inc, Malvern,, PA, United States, ⁴Siemens Healthcare GmbH, Erlangen, Germany, ⁵Radiology and Imaging Sciences, Clinical Center, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, United States

Lung MRI using a high-performance 0.55T system offers reduced susceptibility artifacts, but high resolution imaging is challenged by low proton density, reduced signal-to-noise ratio, and concomitant-fields. 3D spiral acquisitions can be used for improved efficiency and SNR, at the expense of concomitant field related blurring. We present a self-gated, ultra-short echo time, stack-of-spirals acquisition with an efficient inline iterative reconstruction (<15 min) that incorporates both concomitant-field and motion-correction to enable robust high-resolution lung imaging at 0.55T. Our optimized and efficient imaging method is applied to improve image quality in healthy volunteers, patients with lung nodules, and patients with lymphangioleiomyomatosis (LAM).

Hyeonha Kim¹, Seokwon Lee², Jinil Park³, and Jang-Yeon Park^1,2
1Department of Intelligent Precision Healthcare Convergence, Sungkyunkwan University, Suwan, Korea, Republic of, ²Department of Biomedical Engineering, Sungkyunkwan University, Suwan, Korea, Republic of, ³Biomedical Institute for Convergence at SKKU, Sungkyunkwan University, Suwon, Korea, Republic of

Lung MRI is getting more interest as an alternative to CT because of no radiation exposure and expands its role of providing structural information to functional information such as ventilation and perfusion. In terms of 3D functional imaging, 3D ventilation mapping was already proposed, but few studies have been done on 3D perfusion mapping. Here, we propose a 3D pulmonary perfusion map using 3D UTE-MRI with retrospective respiratory and cardiac gating. The proposed method provides high-resolution 3D regional perfusion information of the lungs and will be useful for diagnosing diffusive lung diseases along with ventilation map (e.g., V/Q ratio).

Zachary Peggs¹, Jonathan Brooke², Andrew Cooper¹, Charlotte Bolton², Ian Hall², Susan Francis¹, and Penny Gowland^1
1Sir Peter Mansfield Imaging Centre, University of Nottingham, Nottingham, United Kingdom, ²The Nottingham Biomedical Research Centre (BRC), Respiratory Medicine Dept, School of Medicine, University of Nottingham, Nottingham, United Kingdom

The nature of MR signal variations during healthy ventilation at 3T were explored in 8 healthy participants. Fourier decomposition (FD) techniques were used to remove cardiac signal contributions, and the phase, correlation coefficient and gradient of the parenchyma signal with respect to diaphragm signal were found. This analysis showed that for some voxels, the relationship between parenchyma signal and diaphragm signal is distinctly non-linear. Also, that breathing pattern is an important confounding factor when using the diaphragmatic movement as a surrogate for the change in lung volume, where an additional measure such as a separate 3D acquisition may prove beneficial.

Tahmina Achekzai¹, Luis Loza¹, Stephen Kadlecek¹, Kai Ruppert¹, Faraz Amzajerdian¹, and Rahim Rizi^1
1University of Pennsylvania, Philadelphia, PA, United States

HP-Xe Xenon Transfer Contrast (XTC) imaging produces exchange maps between gas (GP) and red blood cell (RBC) and tissue plasma (TP). However, it is difficult to separate the effects of exchange between RBC or TP due to unintended depolarization by the saturation pulse. We introduce a correction method to  isolate exchange due to either compartment. The correction method measures depolarization due to saturation at a distance Δ, —the separation between the DP resonances—away from the compartment’s resonance frequency. The corrected maps generated here show that greater correction is needed for discrete pulses and higher power pulses. 

Filip Klimeš^1,2, Andreas Voskrebenzev^1,2, Marcel Gutberlet^1,2, Cristian Crisosto^1,2, Robert Grimm³, Frank Wacker^1,2, and Jens Vogel-Claussen^1,2
1Institute of Diagnostic and Interventional Radiology, Hannover Medical School, Hannover, Germany, ²Biomedical Research in Endstage and Obstructive Lung Disease Hannover (BREATH), Member of the German Centre for Lung Research (DZL), Hannover, Germany, ³MR Application Predevelopment, Siemens Healthcare GmbH, Erlangen, Germany

Demonstration of a high repeatability is a mandatory step prior to the usage of new methods in clinical routine. Good to moderate repeatability of 3D phase-resolved functional lung (PREFUL) MRI was shown before. In this study, group-oriented registration (GOREG) was compared to stepwise registration regarding the repeatability of 3D PREFUL ventilation parameters in 30 study participants. Furthermore, image sharpness and structural similarity (SSIM) were assessed. An increased repeatability of dynamic 3D PREFUL ventilation parameters was observed using the GOREG approach. Also, image sharpness and SSIM values were significantly higher for measurements assessed with GOREG registration.

Lea Behrendt^1,2, Marcel Gutberlet^1,2, Andreas Voskrebenzev^1,2, Filip Klimeš^1,2, Arnd Obert^1,2, Agilo Luitger Kern^1,2, Frank Wacker^1,2, and Jens Vogel-Claussen^1,2
1Institute of Diagnostic and Interventional Radiology, Hannover Medical School, Hannover, Germany, ²Biomedical Research in Endstage and Obstructive Lung Disease Hannover (BREATH), German Center for Lung Research (DZL), Hannover, Germany

Phase-resolved Functional Lung imaging (PREFUL) allows the contrast-free quantification of pulmonary ventilation and perfusion dynamics in free breathing. Typically, it is used in conjunction with cartesian spoiled gradient echo sequence (SPGRE) with asymmetric echo. Due to the short T2* times in lung tissue, acquiring signal with echo times as low as possible is essential. Therefore, a 2D ultrashort echo time sequence with a TE of 0.07ms was developed and tested in the PREFUL context in one healthy volunteer. A similar visual impression of derived ventilation maps was found with UTE and SPGRE. However, ventilation defects were more pronounced for UTE.