Synopsis

Keywords: Lung, Contrast Mechanisms, oxygen-enhanced MRI

A recently developed protocol with a dual echo RF-spoiled gradient echo acquisition enabled simultaneous measurement of dynamic OE signal change and the substantial ∆R₂* effect at 3 tesla. To progress towards clinical translation, we evaluated repeatability and reproducibility across two sites, vendors and time points. Our results demonstrate good repeatability for the quantitative and semi-quantitative indices across time points, consistent signal behaviour and repeatable ∆R₂* across two sites and vendors, suggesting potential utility in multi-centre clinical studies. However, echo time dependence should be considered when interpreting percentage signal enhancement.

Introduction

Oxygen-enhanced MRI (OE-MRI) can provide information on regional lung function [1]. Despite technical challenges of OE-MRI in the lung at higher magnetic field, we have recently shown the feasibility of lung OE-MRI at 3 T and signal enhancement behaviour in the lung using a dual echo RF-spoiled gradient echo acquisition [2]. This method enables measurement of dynamic OE signal change with either T₁-weighting or T₂*-weighting, and for monitoring of dynamic ∆R₂* [2,3]. To translate these biomarkers into clinical use, they must demonstrate their repeatable and reproducible assessment of lung function. The purpose of this work was to evaluate the proposed method across two sites, vendors and time points.

Subjects & Methods

Following research ethics approval and written informed consent, we recruited 16 healthy volunteers with no previous record of lung diseases. Of these, 8 travelling subjects (3 males, age range = 26-54 years, median = 39.5) underwent lung MRI on a 3 T whole-body scanner from two vendors in different cities (Philips Ingenia in London, UK and Siemens MAGNETOM Vida in Manchester, UK) at a 4 week interval. Eight separate subjects (4 males, age range = 23-51 years, median = 27) were scanned twice for a scan-rescan test at a 4-6 week interval using the Philips scanner. Scan parameters for the RF-spoiled dual gradient echo acquisition for each vendor are listed in Table 1. Selection of flip angle (FA) was optimized through simulation (Fig. 1) by varying FA and TE values [2,4]. The shortest TE values available for the chosen acquisition were selected for each scanner: TE_1A = 0.71 ms, TE_2A = 1.2 ms for Philips; TE_1B = 0.81 ms, TE_2B = 1.51 ms for Siemens. Subjects breathed medical air (21% O₂) and 100% O₂ via non-rebreathing face mask (Intersurgical Ltd., Wokingham, UK) at a flow rate of 15 Lmin^-1. Images were acquired during free breathing.
Data analysis was performed in MATLAB R2022b (MathWorks, Natick, MA). After non-linear image registration using ANTs [5,6], the lung parenchyma, excluding central major vasculature, was manually segmented from registered images and a voxel-wise tissue density correction was applied. Subsequently, the dynamic data points in the masked lung were fitted using exponential functions for the O₂ wash-in and wash-out with least-squares algorithm for each voxel. Percent signal enhancement (PSE) maps were produced by the subtraction of the fitted baseline normoxia intensity from the fitted hyperoxia intensity, normalized to the fitted normoxia intensity. Mean and standard deviation (std) were calculated for PSE values at each TE over 6 slices for TE_1A, TE_1B, TE_2A and 2 most posterior slices only for TE_2B due to poor signal to noise ratio (SNR) in the anterior slices at TE_2B. Mean ∆R₂* maps were calculated by the subtraction of averaged normoxia R₂* maps (30^th to 60^th time series acquisitions) from averaged hyperoxia R₂* maps (120^th to 180^th). Intra-class correlation coefficient (ICC) was calculated using SPSS v28.0 (SPSS Inc, Chicago, IL).

Results

Simulations show that T₂*-induced negative enhancement (∆S) is maximum at FA ~5° independent of the choice of TE (Fig. 1A), and TE dependence of ∆S reduces with smaller FA (Fig. 1B). Negative PSE increases with lower FA and longer TE (Fig. 1C).
Representative volunteer PSE maps at TE_1A (Fig 2A), TE_2A (Fig. 2B) and △R₂* (Fig. 2C) demonstrate good intra-scanner, intra-subject repeatability, confirmed by Bland-Altman analysis (Fig. 2D-F) (Philips in London). PSE maps from inter-scanner, intra-subject repeat scans show similar spatial distribution of the enhancement signal at TE₁ while the signal at TE₂ from the Manchester site is compromised due to low SNR caused by longer TE (Fig. 3A). Plots of the combined PSE values at 4 separate TEs from two MRI systems display TE dependence of the signal (Fig. 3B, Table 2) similar to simulation expectations. The inter-scanner variability (± std of the parameter differences between two scanners when the same volunteer was scanned) and the intra-scanner variability (between two sessions when the same volunteer was scanned) show good repeatability and reproducibility of the indices (Table 2), confirmed by △R₂* maps and Bland-Altman analysis (Fig. 3C, 3D). High ICC values are observed for both inter-scanner and intra-scanner △R₂* comparisons (Table 2).

Discussion

Our results show that the proposed protocol with a dual echo RF-spoiled gradient echo acquisition at 3 T yields excellent intra-scanner repeatability. They also demonstrate that comparable OE-MRI protocols of the lung can be implemented at 3 T across different sites and scanners with good repeatability and reproducibility for △R₂*. Although matching TE between scanners is challenging, the expected trend of PSE with TE is observed between sites.

Conclusions

We have evaluated repeatability and reproducibility of OE-MRI of the lung at 3 T across two sites, vendors and time points using a dual echo RF-spoiled gradient echo acquisition. Our results demonstrate good repeatability for the quantitative and semi-quantitative indices across time points, consistent signal behaviour and repeatable ∆R₂* across two sites and vendors, suggesting potential utility in multi-centre clinical studies. However, substantial impact of TE should be considered when interpretating PSE.

Acknowledgements

This work is supported by the Cancer Research UK National Cancer Imaging Translational Accelerator (NCITA) awards C1519/A28682 (UCL) and C19221/A28683 (University of Manchester), the EPSRC-funded UCL Centre for Doctoral Training in Medical Imaging (EP/L016478/1; EP/S021930/1), an EPSRC Industrial CASE award (Voucher No. V20000074), GlaxoSmithKline Research and Development Ltd (BIDS3000035683), and Innovate UK award 104629. We thank Lucy Caselton, Sumandeep Kaur and David M. Higgins for the technical assistance.