Rapid single‐breath hyperpolarized noble gas MRI‐based biomarkers of airspace enlargement

Abstract

Background Multi‐b diffusion‐weighted hyperpolarized‐gas MRI measures pulmonary airspace‐enlargement using apparent diffusion coefficients (ADCs) and mean‐linear‐intercepts (Lm). Purpose To develop single‐breath 3D multi‐b diffusion‐weighted 3He and 129Xe MRI using k‐space undersampling. Rapid, cost‐efficient, single‐breath acquisitions may facilitate clinical translation. Study Type Prospective. Subjects We evaluated 12 participants, including nine subjects (mean age = 69 ± 9) who were included in the retrospective experiment and three chronic pulmonary obstruction disease (COPD) patients (mean age = 81 ± 6) who participated in the prospective study. Field Strength A whole‐body 3 T 2D/3D fast gradient recall echo (FGRE) sequence. Assessment Hyperpolarized 3He/129Xe MRI, spirometry, plethysmography computed tomography (CT). We evaluated 129Xe ADC/morphometry estimates by retrospectively undersampling previously acquired fully sampled multibreath, multi‐b diffusion‐weighted data. Next, we prospectively evaluated the feasibility of accelerated (AF = 7) 3He MRI static‐ventilation/T2* (extra short‐TE, b = 0 image) and ADC/morphometry (five b‐values) maps using a single gas‐dose and 16‐second breath‐hold. To conservatively evaluate cost‐improvement, we compared total costs of single vs. multiple 129Xe doses. Statistical Tests Multivariate analysis of variance, independent t‐tests and voxel‐by‐voxel basis difference test. Results For the retrospectively undersampled 129Xe data, a nonsignificant mean difference for ADC/Lm of 14%/12%, 12%/8%, and 11%/9% was observed (all, P > 0.4) between the fully sampled and accelerated data for the never‐smoker, COPD, and alpha‐1 antitrypsin deficiency (AATD) groups, respectively. The control never‐smoker group had significantly lower ADC (P < 0.001) and Lm (P < 0.001) than the COPD/AATD group for both fully sampled and accelerated data. For the prospectively acquired 3He MRI data, static‐ventilation, T2*, ADC, and morphometry maps were acquired using a single 16‐second breath‐hold scan and single gas dose. Accelerated imaging resulted in cost savings of ~$US 1000/patient, a conservative estimate based on 129Xe MRI dose savings (single vs. five doses). Data Conclusion This is a proof‐of‐concept demonstration of accelerated (7×) morphometry that shows that less cost‐ and time‐efficient multibreath methods that lead to variability and patient fatigue may be avoided in the future. Level of Evidence: 2 Technical Efficacy: Stage 5 J. Magn. Reson. Imaging 2018.

Assessment of pulmonary morphometry using hyperpolarized 129Xe diffusion‐weighted MRI with variable‐sampling‐ratio compressed sensing patterns

“…25 Meanwhile, in our study, the means of SNR of images with maximum b-value of 30 s/cm 2 were 16, 23, and 21 for FS, rVCS, and rCS measurements, respectively. The lowest SNR was 13, which met the Rose criteria of SNR = 5 for the lung morphometric estimates, 39,40 and the measured SNRs for each subject are summarized in Table S4. In addition, our retrospective results suggested that the proposed method can achieve comparable image quality (MAE < 0.03) 25 and reliable lung morphological parameters (the differences were less than 3%) with the FS acquisition.…”

Accelerate gas diffusion-weighted MRI for lung morphometry with deep learning

“…2 ). Nevertheless, the reconstruction of SIDER needs pre-estimated model parameters provided by the TV reconstructed results, resulting in long reconstruction time [ 12 ]. Although TV and SIDER can remove aliasing artifacts to some extent, they lead to over-smooth appearances in lung regions owing to the high acceleration factor.…”

“…Successful application of deep learning relies on large and high-quality training data, which may not be feasible for some medical imaging problems [ 40 ]. This is especially true for multiple b -value gas DW-MRI due to the limitations of non-renewable gas hyperpolarization and breath-holding duration [ 12 ]. Training on simulated data is one of the strategies to eliminate this issue and permit generating large amounts of training data with diverse experimental conditions [ 41 , 42 ].…”

“…Furthermore, multiple b -value gas DW-MRI, together with theoretical diffusion models (e.g., cylinder model [ 8 , 9 ] or stretched exponential model [ 10 ]), enables non-invasive and quantitative estimation of lung morphometry parameters at the alveolar level [ 11 ]. However, such technique suffers from long scan times due to multiple b -value acquisitions and long diffusion gradient time [ 12 ], which imposes great burdens on subjects, especially for children and patients with lung diseases.…”

Pulmonary delivery of nano‐particles for lung cancer diagnosis and therapy: Recent advances and future prospects

“…Second, most hyperpolarized gas MRI methods require completion within a single closed breath, which can impose a considerable burden on patients due to the extended holding time. Third, the achievement of targeted modification and visualization of various molecular targets of lung cancer using hyperpolarized 129 Xe gas poses significant challenges (Abascal et al, 2018; Duan et al, 2019; Ireland et al, 2010; Westcott et al, 2018; Xiao et al, 2019). In comparison to hyperpolarized gas imaging, the utilization of nano‐particles as an intelligent and multifunctional platform for biomedical applications presents significant advantages, demonstrated by their targeted delivery of antitumor drugs, efficient capacity for drug‐loading, controllable release, and among other notable characteristics.…”