0442

Binning method for treatment response mapping with hyperpolarized gas lung MRI: application in subjects with asthma

Guilhem Jean Collier¹, Alberto Biancardi¹, Paul J Hughes¹, Laurie Smith¹, Grace T Mussel¹, Helen Marshall¹, Ho F Chan¹, Graham Norquay¹, and Jim M Wild¹
¹Department of Infection, Immunity & Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom

Synopsis

This work applies a clustering method developed for image analysis of hyperpolarised gas lung ventilation MRI to a cohort of patients referred from a severe asthma clinic for investigation of breathlessness. Patients underwent spirometry tests and imaging at baseline and after inhalation of Salbutamol to assess reversibility. In a subset of patients, images pre and post bronchodilator were registered and a novel treatment response mapping method was applied. Results show significant correlations at baseline between imaging markers and FEV1% and between their percentage changes and the reversibility testing. The treatment response mapping method offers additional robust additional insights.

Introduction

Pulmonary MR imaging with hyperpolarized (HP) ³He and ¹²⁹Xe enables the direct visualization of gas distribution in the lungs and has been used to study the heterogeneity of regional ventilation and response to therapy in patients with asthma^1,2. In an effort to quantify the ventilation distribution in its entirety, a clustering method producing linear binning maps has been proposed^3,4. This method offers the advantage of defining not only ventilation defect regions (VDR: regions with no gas signal) but also regions of low ventilation (LVR), normal ventilation (NVR) and hyper ventilation (HVR). An alternative normalization method has been proposed that increases the reproducibility of ventilation categorization⁵. A treatment response mapping (TRM) method for HP gas imaging had been previously developed⁶, based on a strictly arithmetic approach, to study changes pre and post bronchodilator (BD) in patients with asthma. To overcome the difficulty in interpreting changes involving hyperventilated regions, this study proposes a novel TRM method that analyses the transitions between binning categories. It is applied to reversibility testing in patients with asthma.

Methods

24 subjects referred from a severe asthma clinic for investigation of breathlessness were assessed with HP ¹²⁹Xe MRI and spirometry at baseline and approximately 20 minutes after inhalation of 400μg of salbutamol. Prior to testing, patients were advised to withhold all inhalers on the morning of the scan. All imaging protocols were performed on a 1.5T GE HDx MR scanner. A dose of approximately 500mL of HP ¹²⁹Xe at ~ 30% polarisation mixed with N₂ up to 1L was used for ventilation imaging (3D steady state free precession ¹²⁹Xe⁷ and spatially-matched ¹H anatomical images). The binning method implemented has been described by Collier et al5. Briefly, the images are first corrected for bias field and normalized by the average ¹²⁹Xe signal in the thoracic cavity. Threshold values previously optimised in cohorts of healthy volunteers are used to define different regions of ventilation (Fig. 1). Binning maps and histogram distribution of signal intensity were produced for each image. Percentage ventilation in the different bins and a global coefficient of variation (CV) of ventilation were derived. In a subset of 8 patients, an additional registration⁸ between baseline and post bronchodilator images was performed. Using a defined set of rules (Fig. 4.e), the intensity changes for each pixel were displayed in an outcome treatment response map as improvement, deterioration or irrelevant change. By summing the pertinent changes over the entire lung, a treatment response parameter TR expressed in mL of lung ventilation improvement was derived. TR was also expressed as a percentage of total thoracic cavity volume TR%. Spearman correlations at baseline between imaging parameters and FEV1% and correlations between percentage changes in FEV1 and imaging parameters were computed.

Results

Of the 24 patients assessed, there was a significant (p<0.001) increase in the % change in FEV1 (median difference (MD)=7.1%) and decrease in the % change in VDR (MD=-29.1%) post salbutamol. According to BTS SIGN asthma guidelines⁹, 13/24 patients had a clinically significant change in FEV1. In these 13 patients, the MD in FEV1=21.6% and the MD in VDR=-38%. This change in FEV1 and VDR was significantly larger (p<0.001) when compared to those who did not meet BTS SIGN guidelines (11 patients). In these 11 patients, the MD for VDR=-12.6%, when compared to 4.6% for FEV1. On average the relative change in VDR is approximately double the relative change in FEV1. Example images, binning maps and histograms for a responder and a non-responder are displayed on figures 2 and 3 respectively. At baseline, there was a good correlation between FEV1% predicted, VDR and CV (Fig. 5.a & 5.b). The % change in FEV1 was significantly correlated with % change in VDR (Fig. 5.c) and ΔCV (r=-0.64, p=0.0008). In the subgroup of 8 subjects however, no significant correlation was found for both VDR or TR% (Fig. 5.d). Example TRM and comparison with previous TRM method is displayed on figure 4.

Discussion

The yellow arrows in Fig. 4 highlight the main advantage of this novel method. By defining a normal or sufficient range of ventilation as opposed to abnormal (VDR/LVR), areas where signal intensity decreases from HVR to NVR that were previously⁶ negatively contributing toward a global ventilation change parameter can now be discarded from the TR calculation in the proposed method and displayed with a different colour map. Although the % change in FEV1 did not significantly correlate with TR%, it also did not correlate with the % change in VDR in these 8 subjects whereas the correlation was strong when considering the 24 patients. Future work will focus on applying the TRM method to all subjects. FEV1 measurement has also some limitations and in the 11 patients without a significant change in FEV1, changes in ventilation were visible for some patients post Salbutamol.

Conclusions

VDR, CV and their relative changes calculated using the binning method show strong correlation with the clinical metric of FEV1 in patients with asthma. The novel binning-based TRM proposed method provides additional regional information that may help the clinicians to determine which patients have a significant response to bronchodilators and aid in the diagnosis of asthma.

Acknowledgements

This work was supported by NIHR grant NIHR-RP-R3-12-027 and MRC grant MR/M008894/1. The views expressed in this work are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health.

References

[1] Altes, T. et al. Hyperpolarized 3He MR lung ventilation imaging in asthmatics: preliminary findings. J Magn Reson Imaging. 2001;13(3):378–84.

[2] Svenningsen, S. et al. Hyperpolarized 3He and 129Xe MRI: Differences in asthma before bronchodilation. J Magn Reson Imaging. 2013;38(6):1521–30.

[3] He, M. et al. Extending semiautomatic ventilation defect analysis for hyperpolarized (129)Xe ventilation MRI. Acad Radiol 21, 1530-1541, doi:10.1016/j.acra.2014.07.017 (2014).

[4] He, M. et al. Using Hyperpolarized 129Xe MRI to Quantify the Pulmonary Ventilation Distribution. Acad Radiol 23, 1521-1531 (2016).

[5] Collier, G. et al. ISMRM abstract 4482, 2018.

[6] Horn, F. et al. Regional Ventilation Changes in the Lung: Treatment Response Mapping by Using Hyperpolarized Gas MR Imaging as a Quantitative Biomarker. Radiology. 2017;284(3):854-861

[7] Stewart, N. et al. Feasibility of human lung ventilation imaging using highly polarized naturally abundant xenon and optimized three‐dimensional steady‐state free precession. Magn Reson Med. 2015;74:346-352.

[8] Barber, D. et al. Automatic segmentation of medical images using image registration: diagnostic and simulation applications. J Med Eng Technol. 2005;29(2):5363.

[9] BTS/SIGN British Guideline on the Management of Asthma. 2019. Available from: https://www.brit-thoracic.org.uk/quality-improvement/guidelines/asthma/

Figures

Figure 1: a) Histogram distribution of normalised ventilation signal intensity in a healthy volunteer with the 6 bins defining regions of ventilation defect VDR [0, 0.33], low ventilation LVR [0.33, 0.66], normal ventilation NVR [0.66, 1.33] and hyper ventilation HVR [1.33, max]. b) Representative coronal slice of HP gas distribution in this subject and corresponding binning map c). Example of histogram distribution (d), ventilation coronal image (e) and corresponding binning map (f) for a patient with chronic obstructive pulmonary disease.

Figure 2: Example of a subject who responded to inhalation of Salbutamol (FEV1% predicted at baseline: 55.5%, post BD: 72.8%). HP ¹²⁹Xe ventilation MR images at baseline (a) with corresponding binning maps (b), VDR=30.4%, LVR=14.9%, CV=0.86) and histogram distribution of normalized signal intensity (c). Post BD images (d), binning maps (e), VDR=17.6%, LVR=18.1%, CV=0.64) and histogram (f) after inhalation of Salbutamol.

Figure 3: Example of a non-responder subject (FEV1% predicted at baseline: 109.9%, post BD: 116.7%). HP 129Xe ventilation MR Images at baseline (a) with corresponding binning maps (b), VDR=2.9%, LVR=14.2%, CV=0.33) and histogram distribution of normalized signal intensity (c). Post BD images (d), binning maps (e), VDR=2.0%, LVR=12.2%, CV=0.31) and histogram (f) after inhalation of Salbutamol. Although the % change in FEV1 is non-significant (6.2%), a visual improvement of the histogram shape can be noticed and a decrease of 2.9% in abnormal ventilation (VDR/LVR) measured.

Figure 4: Example of TRM in a subject with asthma: a) baseline images, b) post BD images, c) TRM as performed as in Horn et al⁶, d) proposed novel TRM method. The arrows on top of the reference healthy histogram in e) represent the different rules used to create the maps. A pixel is displayed in level of green if pre/post BD change is positive and pre BD value belongs to VDR/LVR (<0.66). If the change is negative and post BD value < 0.66, it is displayed in level of red. On contrary, it is displayed in white if post BD value > 0.66, counted as non-pertinent change and discarded during the TR calculation.

Figure 5: Baseline Spearman correlations between FEV1% predicted and VDR (a) and CV (b). c) Spearman correlation between percentage change in FEV1 (between baseline and post BD) and % change in VDR. d) Spearman correlation between % change in FEV1 and TR%.

Proc. Intl. Soc. Mag. Reson. Med. 28 (2020)

0442