Accelerated ¹²⁹Xe MRI morphometry of terminal airspace enlargement: Feasibility in volunteers and those with alpha‐1 antitrypsin deficiency

Abstract

Purpose Multi‐b diffusion‐weighted hyperpolarized inhaled‐gas MRI provides imaging biomarkers of terminal airspace enlargement including ADC and mean linear intercept (Lm), but clinical translation has been limited because image acquisition requires relatively long or multiple breath‐holds that are not well‐tolerated by patients. Therefore, we aimed to accelerate single breath‐hold 3D multi‐b diffusion‐weighted 129Xe MRI, using k‐space undersampling in imaging direction using a different undersampling pattern for different b‐values combined with the stretched exponential model to generate maps of ventilation, apparent transverse relaxation time constant (T2∗), ADC, and Lm values in a single, short breath‐hold; accelerated and non‐accelerated measurements were directly compared. Methods We evaluated multi‐b (0, 12, 20, 30, and 45.5 s/cm2) diffusion‐weighted 129Xe T2∗/ADC/morphometry estimates using acceleration factor (AF = 1 and 7) and multi‐breath sampling in 3 volunteers (HV), and 6 participants with alpha‐1 antitrypsin deficiency (AATD). Results For the HV subgroup, mean differences of 5%, 2%, and 8% were observed between fully sampled and undersampled k‐space for ADC, Lm, and T2∗ values, respectively. For the AATD subgroup, mean differences were 9%, 6%, and 12% between fully sampled and undersampled k‐space for ADC, Lm and T2∗ values, respectively. Although mean differences of 1% and 4.5% were observed between accelerated and multi‐breath sampled ADC and Lm values, respectively, mean ADC/Lm estimates were not significantly different from corresponding mean ADCM/LmM or mean ADCA/LmA estimates (all P > 0.60 , A = undersampled and M = multi‐breath sampled). Conclusions Accelerated multi‐b diffusion‐weighted 129Xe MRI is feasible at AF = 7 for generating pulmonary ADC and Lm in AATD and normal lung.

Protocols for multi‐site trials using hyperpolarized 129Xe MRI for imaging of ventilation, alveolar‐airspace size, and gas exchange: A position paper from the 129Xe MRI clinical trials consortium

“…Increases in encoding efficiency, such as using spiral trajectories or a standardized compressed sensing method, would be highly beneficial to alveolar-airspace size imaging. 61,62,64,96 Additionally, analysis should be standardized. Basic mapping of ADC is relatively straightforward using Equation ( 3), but more advanced diffusion morphometry analysis requires multiple images with different diffusion weighting.…”

“…Most commonly, the total volume of gas delivered to subjects (usually 1 L) has been held constant for all subjects independent of age and lung size. 20,24,31,48,62,83 However, others have tailored total volume based on a subject's total lung capacity (TLC) 16,38 or forced vital capacity (FVC). 10,63 To date, a consensus has yet to be reached regarding best practices for calculating dose volumes.…”

“…19,[84][85][86] Lung inflation during HP gas MRI is determined both by the initial inflation state of the lungs prior to xenon inhalation and by the total volume of gas delivered to the subject. To date, the majority of studies have had subjects inhale doses from functional residual capacity (FRC), 1,38,48,59,62,83,87,88 though some have started from residual volume (RV). 58,63 Ultimately, it is unclear whether inhaling from FRC or RV provides better reproducibility of lung inflation.…”

“…Alveolar-airspace size imaging is most often performed using a diffusion weighted, 2D slice selective gradient echo sequence with bipolar diffusion gradients. 24,26,60,62,63 For applications requiring only ADC, it is common to use two diffusion weightings (b-values), often b = 0, and 12 s/cm 2 . 24,26 ADC is calculated using where S 0 is the signal without diffusion weighting and S b is that for the image acquired with b-value b.…”

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

“…Compressed sensing MRI has been applied to accelerate hyperpolarized gas DW-MRI acquisition [11][12][13][14][15][16], because it can reduce scan time by undersampling kspace without additional hardware requirement [17][18][19]. However, compressed sensing MRI generally uses fixed sparsifying transforms, which may not capture complex image characteristics associated with biological tissues [20,21].…”