Extending Semiautomatic Ventilation Defect Analysis for Hyperpolarized 129Xe Ventilation MRI

doi:10.1016/j.acra.2014.07.017

Academic Radiology

Volume 21, Issue 12, December 2014, Pages 1530-1541

https://doi.org/10.1016/j.acra.2014.07.017 Get rights and content

Rationale and Objectives

Clinical deployment of hyperpolarized ¹²⁹Xe magnetic resonance imaging requires accurate quantification and visualization of the ventilation defect percentage (VDP). Here, we improve the robustness of our previous semiautomated analysis method to reduce operator dependence, correct for B₁ inhomogeneity and vascular structures, and extend the analysis to display multiple intensity clusters.

Materials and Methods

Two segmentation methods were compared—a seeded region-growing method, previously validated by expert reader scoring, and a new linear-binning method that corrects the effects of bias field and vascular structures. The new method removes nearly all operator interventions by rescaling the ¹²⁹Xe magnetic resonance images to the 99th percentile of the cumulative distribution and applying fixed thresholds to classify ¹²⁹Xe voxels into four clusters: defect, low, medium, and high intensity. The methods were applied to 24 subjects including patients with chronic obstructive pulmonary disease (n = 8), age-matched controls (n = 8), and healthy normal subjects (n = 8).

Results

Linear-binning enabled a faster and more reproducible workflow and permitted analysis of an additional 0.25 ± 0.18 L of lung volume by accounting for vasculature. Like region-growing, linear-binning VDP correlated strongly with reader scoring (R² = 0.93, P < .0001), but with less systematic bias. Moreover, linear-binning maps clearly depict regions of low and high intensity that may prove useful for phenotyping subjects with chronic obstructive pulmonary disease.

Conclusions

Corrected linear-binning provides a robust means to quantify ¹²⁹Xe ventilation images yielding VDP values that are indistinguishable from expert reader scores, while exploiting the entire dynamic range to depict multiple image clusters.

Section snippets

Materials and methods

All studies were approved by the Institutional Review Board and before enrolment, written informed consent was obtained from all subjects. Image analysis was conducted using data acquired during a previously reported clinical trial of HP ¹²⁹Xe (6), and hence the acquisition methods are only briefly summarized here. To test the new analysis methods, a total of 24 ¹²⁹Xe ventilation images were analyzed. This included eight young healthy volunteers (HVs; six female, two male; mean age

Linear-Binning Histogram Scaling

Figure 4 illustrates the importance of correctly handling the high-intensity tail (Fig 4d) in the ¹²⁹Xe intensity histogram before rescaling to the range of 0–1. A previous approach we had suggested (6) was to divide all intensities by the average of the top 5% of values. As illustrated in Figure 4e, this scaling does reduce the tail, but does not completely remove it. The rescaled histogram is still significantly weighted toward low-intensity values and the resulting maps significantly

Advantages of the Corrected Linear-Binning Method

The corrected linear-binning method provides an elegant way to quantify and visualize ventilation defects and has the capability to highlight subtle features of the ventilation distribution that could be missed by the simple binary classification. Moreover, linear-binning mitigates the subjective elements of the analysis caused by intra- and interobserver bias while providing a substantial throughput improvement. The only user intervention in the process is initialization and supervision of the

Conclusions

This study demonstrated the feasibility of using corrected linear-binning analysis of ¹²⁹Xe MRI to visualize and quantify regional ventilation in HVs, AMCs, and patients with COPD. The resulting defect percentages correlated exceptionally well and were of similar magnitude to expert reader scores, suggesting that linear binning closely follows expert reader observations. Moreover, this linear-binning method provides a promising means to not only accelerate objective image analysis, but also by

Acknowledgments

The authors would like to thank Sally Zimney for careful proofreading of the manuscript.

References (22)

Y. Shukla et al.
Hyperpolarized 129Xe magnetic resonance imaging: tolerability in healthy volunteers and subjects with pulmonary disease
Acad Radiol
(2012)
M. Kirby et al.
Hyperpolarized 3He magnetic resonance functional imaging semiautomated segmentation
Acad Radiol
(2012)
C. Studholme et al.
An overlap invariant entropy measure of 3D medical image alignment
Pattern Recogn
(1999)
B. Belaroussi et al.
Intensity non-uniformity correction in MRI: existing methods and their validation
Med Image Anal
(2006)
J.B. Arnold et al.
Qualitative and quantitative evaluation of six algorithms for correcting intensity nonuniformity effects
Neuroimage
(2001)
S. Svenningsen et al.
Hyperpolarized He and Xe MRI: differences in asthma before bronchodilation
J Magn Reson Imaging
(2013)
S.S. Kaushik et al.
Probing the regional distribution of pulmonary gas-exchange through single-breath, gas- and dissolved-phase 129Xe MR imaging
J Appl Physiol
(2013)
K. Qing et al.
Regional mapping of gas uptake by blood and tissue in the human lung using hyperpolarized xenon-129 MRI
J Magn Reson Imaging
(2014)
Z.I. Cleveland et al.
Hyperpolarized Xe MR imaging of alveolar gas uptake in humans
PLoS One
(2010)
D. Kramer
DOE begins rationing helium-3
Phys Today
(2010)

R.S. Virgincar et al.

Quantitative analysis of hyperpolarized (129) Xe ventilation imaging in healthy volunteers and subjects with chronic obstructive pulmonary disease

NMR Biomed

(2012)

Cited by (74)

Hyperpolarized 129Xe MRI, 99mTc scintigraphy, and SPECT in lung ventilation imaging: a quantitative comparison
2024, Academic Radiology
The current clinical standard for functional imaging of patients with lung ailments is nuclear medicine scintigraphy and Single Photon Emission Computed Tomography (SPECT) which detect the gamma decay of inhaled radioactive tracers. Hyperpolarized (HP) Xenon-129 MRI (XeMRI) of the lungs has recently been FDA approved and provides similar functional images of the lungs with higher spatial resolution than scintigraphy and SPECT. Here we compare Technetium-99m (^99mTc) diethylene-triamine-pentaacetate scintigraphy and SPECT with HP XeMRI in healthy controls, asthma, and chronic obstructive pulmonary disorder (COPD) patients.
59 subjects, healthy, with asthma, and with COPD, underwent ^99mTc scintigraphy/SPECT, standard spirometry, and HP XeMRI. XeMRI and SPECT images were registered for direct voxel-wise signal comparisons. Images were also compared using ventilation defect percentage (VDP), and a standard 6-compartment method. VDP calculated from XeMRI and SPECT images was compared to spirometry.
Median Pearson correlation coefficient for voxel-wise signal comparison was 0.698 (0.613–0.782) between scintigraphy and XeMRI and 0.398 (0.286–0.502) between SPECT and XeMRI. Correlation between VDP measures was r = 0.853, p < 0.05. VDP separated asthma and COPD from the control group and was significantly correlated with FEV1, FEV1/FVC, and FEF 25–75.
HP XeMRI provides equivalent information to ^99mTc SPECT and standard spirometry measures. Additionally, XeMRI is non-invasive, hence it could be used for longitudinal studies for evaluating emerging treatment for lung ailments.
129Xe MRI Ventilation Defects in Asthma: What is the Upper Limit of Normal and Minimal Clinically Important Difference?
2023, Academic Radiology
The minimal clinically important difference (MCID) and upper limit of normal (ULN) for MRI ventilation defect percent (VDP) were previously reported for hyperpolarized ³He gas MRI. Hyperpolarized ¹²⁹Xe VDP is more sensitive to airway dysfunction than ³He, therefore the objective of this study was to determine the ULN and MCID for ¹²⁹Xe MRI VDP in healthy and asthma participants.
We retrospectively evaluated healthy and asthma participants who underwent spirometry and ¹²⁹XeMRI on a single visit; participants with asthma completed the asthma control questionnaire (ACQ-7). The MCID was estimated using distribution- (smallest detectable difference [SDD]) and anchor-based (ACQ-7) methods. Two observers measured VDP (semiautomated k-means-cluster segmentation algorithm) in 10 participants with asthma, five-times each in random order, to determine SDD. The ULN was estimated based on the 95% confidence interval of the relationships between VDP and age.
Mean VDP was 1.6 ± 1.2% for healthy (n = 27) and 13.7 ± 12.9% for asthma participants (n = 55). ACQ-7 and VDP were correlated (r = .37, p = .006; VDP = 3.5·ACQ + 4.9). The anchor-based MCID was 1.75% while the mean SDD and distribution-based MCID was 2.25%. VDP was correlated with age for healthy participants (p = .56, p =.003; VDP = .04·Age-.01). The ULN for all healthy participants was 2.0%. By age tertiles, the ULN was 1.3% ages 18–39 years, 2.5% for 40–59 years and 3.8% for 60–79 years.
The ¹²⁹Xe MRI VDP MCID was estimated in participants with asthma; the ULN was estimated in healthy participants across a range of ages, both of which provide a way to interpret VDP measurements in clinical investigations.
Combining neural networks and image synthesis to enable automatic thoracic cavity segmentation of hyperpolarized 129Xe MRI without proton scans
2023, Magnetic Resonance Imaging
Quantification of ¹²⁹Xe MRI relies on accurate segmentation of the thoracic cavity, typically performed manually using a combination of ¹H and ¹²⁹Xe scans. This can be accelerated by using Convolutional Neural Networks (CNNs) that segment only the ¹²⁹Xe scan. However, this task is complicated by peripheral ventilation defects, which requires training CNNs with large, diverse datasets. Here, we accelerate the creation of training data by synthesizing ¹²⁹Xe images with a variety of defects. We use this to train a 3D model to provide thoracic cavity segmentation from ¹²⁹Xe ventilation MRI alone.
Training and testing data consisted of 22 and 33 3D ¹²⁹Xe ventilation images. Training data were expanded to 484 using Template-based augmentation while an additional 298 images were synthesized using the Pix2Pix model. This data was used to train both a 2D U-net and 3D V-net-based segmentation model using a combination of Dice-Focal and Anatomical Constraint loss functions. Segmentation performance was compared using Dice coefficients calculated over the entire lung and within ventilation defects.
Performance of both U-net and 3D segmentation was improved by including synthetic training data. The 3D models performed significantly better than U-net, and the 3D model trained with synthetic ¹²⁹Xe images exhibited the highest overall Dice score of 0.929. Moreover, addition of synthetic training data improved the Dice score in ventilation defect regions from 0.545 to 0.588 for U-net and 0.739 to 0.765 for the 3D model.
It is feasible to obtain high-quality segmentations from ¹²⁹Xe scan alone using 3D models trained with additional synthetic images.
Hyperpolarized 129Xenon MRI Ventilation Defect Quantification via Thresholding and Linear Binning in Multiple Pulmonary Diseases
2022, Academic Radiology
Citation Excerpt :
A vesselness filter has been developed and utilized by different groups though with variations in the technique (23,30,33,48). Unlike 3He ventilation images with higher SNR and spatial resolution, 129Xe ventilation images require a co-registered proton image set of the thoracic cavity to correctly identity and label vasculature to be removed from the ventilation image (30,33). In our study, an inspiratory level matched thoracic proton image set was not acquired, therefore, we utilized a median filter after manual lung and large vessel segmentation prior to VDP quantification.
There is no agreed upon method for quantifying ventilation defect percentage (VDP) with high sensitivity and specificity from hyperpolarized (HP) gas ventilation MR images in multiple pulmonary diseases for both pediatrics and adults, yet identifying such methods will be necessary for future multi-site trials. Most HP gas MRI ventilation research focuses on a specific pulmonary disease and utilizes one quantification scheme for determining VDP. Here we sought to determine the potential of different methods for quantifying VDP from HP ¹²⁹Xe images in multiple pulmonary diseases through comparison of the most utilized quantification schemes: linear binning and thresholding.
HP ¹²⁹Xe MRI was performed in a total of 176 subjects (125 pediatrics and 51 adults, age 20.98±16.48 years) who were either healthy controls (n = 23) or clinically diagnosed with cystic fibrosis (CF) (n = 37), lymphangioleiomyomatosis (LAM) (n = 29), asthma (n = 22), systemic juvenile idiopathic arthritis (sJIA) (n = 11), interstitial lung disease (ILD) (n = 7), or were bone marrow transplant (BMT) recipients (n = 47). HP ¹²⁹Xe ventilation images were acquired during a ≤16 second breath-hold using a 2D multi-slice gradient echo sequence on a 3T Philips scanner (TR/TE 8.0/4.0ms, FA 10-12°, FOV 300 × 300mm, voxel size≈3 × 3 × 15mm). Images were analyzed using 5 different methods to quantify VDPs: linear binning (histogram normalization with binning into 6 clusters) following either linear or a variant of a nonparametric nonuniform intensity normalization algorithm (N4ITK) bias-field correction, thresholding ≤60% of the mean signal intensity with linear bias-field correction, and thresholding ≤60% and ≤75% of the mean signal intensity following N4ITK bias-field correction. Spirometry was successfully obtained in 84% of subjects.
All quantification schemes were able to label visually identifiable ventilation defects in similar regions within all subjects. The VDPs of control subjects were significantly lower (p<0.05) compared to BMT, CF, LAM, and ILD subjects for most of the quantification methods. No one quantification scheme was better able to differentiate individual disease groups from the control group. Advanced statistical modeling of the VDP quantification schemes revealed that in comparing controls to the combined disease group, N4ITK bias-field corrected 60% thresholding had the highest predictive efficacy, sensitivity, and specificity at the VDP cut-point of 2.3%. However, compared to the thresholding quantification schemes, linear binning was able to capture and label subtle low-ventilation regions in subjects with milder obstruction, such as subjects with asthma.
The difference in VDP between healthy controls and patients varied between the different disease states for all quantification methods. Although N4ITK bias-field corrected 60% thresholding was superior in separating the combined diseased group from controls, linear binning is able to better label low-ventilation regions unlike the current, 60% thresholding scheme. For future clinical trials, a consensus will need to be reached on which VDP scheme to utilize, as there are subtle advantages for each for specific disease.
Comparison of Hyperpolarized 3He and 129Xe MR Imaging in Cystic Fibrosis Patients
2022, Academic Radiology
In this study, we compared hyperpolarized ³He and ¹²⁹Xe images from patients with cystic fibrosis using two commonly applied magnetic resonance sequences, standard gradient echo (GRE) and balanced steady-state free precession (TrueFISP) to quantify regional similarities and differences in signal distribution and defect analysis.
Ten patients (7M/3F) with cystic fibrosis underwent hyperpolarized gas MR imaging with both ³He and ¹²⁹Xe. Six had MRI with both GRE, and TrueFISP sequences and four patients had only GRE sequence but not TrueFISP. Ventilation defect percentages (VDPs) were calculated as lung voxels with <60% of the whole-lung hyperpolarized gas signal mean and was measured in all datasets. The voxel signal distributions of both ¹²⁹Xe and ³He gases were visualized and compared using violin plots. VDPs of hyperpolarized 3 He and 129 Xe were compared in Bland-Altman plots; Pearson correlation coefficients were used to evaluate the relationships between inter-gas and inter-scan to assess the reproducibility.
A significant correlation was demonstrated between ¹²⁹Xe VDP and ³He VDP for both GRE and TrueFISP sequences (ρ = 0.78, p<0.0004). The correlation between the GRE and TrueFISP VDP for ³He was ρ = 0.98 and was ρ = 0.91 for ¹²⁹Xe. Overall, ¹²⁹Xe (27.2±9.4) VDP was higher than ³He (24.3±6.9) VDP on average on cystic fibrosis patients.
In patients with cystic fibrosis, the selection of hyperpolarized ¹²⁹Xe or ³He gas is most likely inconsequential when it comes to measure the overall lung function by VDP although ¹²⁹Xe may be more sensitive to starker lung defects, particularly when using a TrueFISP sequence.
Comparison of Functional Free-Breathing Pulmonary 1H and Hyperpolarized 129Xe Magnetic Resonance Imaging in Pediatric Cystic Fibrosis
2021, Academic Radiology
Citation Excerpt :
This study chose k-means clustering as the technique for calculating VDP, since it is one of the most prevalent techniques used in the HP gas imaging literature (9,13,38). The k-means VDP segmentation that was previously optimized for HP gas MRI (9) was assumed to be valid for PREFUL FV maps; however, future work will investigate alternative VDP calculation methods, such as linear binning (39,40). This method uses healthy reference data to set the bin thresholds (41), which may help to decouple ventilation defects from mechanically stiff lung regions in the signal histogram.
Phase resolved functional lung (PREFUL) magnetic resonance imaging (MRI) is a free-breathing ¹H-based technique that produces maps of fractional ventilation (FV). This study compared ventilation defect percent (VDP) calculated using PREFUL to hyperpolarized (HP) ¹²⁹Xe MRI and pulmonary function tests in pediatric cystic fibrosis (CF).
27 pediatric participants were recruited (mean age 13.0 ± 2.7), including 6 with clinically stable CF, 11 CF patients undergoing a pulmonary exacerbation (PEx), and 10 healthy controls. Spirometry was performed to measure forced expiratory volume in 1 second (FEV₁), along with nitrogen multiple breath washout to measure lung clearance index (LCI). VDP was calculated from single central coronal slice PREFUL FV maps and the corresponding HP ¹²⁹Xe slice.
The stable CF group had a normal FEV₁ (p = 0.41) and elevated LCI (p = 0.007). The CF PEx group had a decreased FEV₁ (p < 0.0001) and elevated LCI (p < 0.0001). PREFUL and HP ¹²⁹Xe VDP were significantly different between the CF PEx and healthy groups (p < 0.05). In the stable CF group, PREFUL and HP ¹²⁹Xe VDP were not significantly different from the healthy group (p = 0.18 and 0.08, respectively). There was a correlation between PREFUL and HP ¹²⁹Xe VDP (R² = 0.31, p = 0.004), and both parameters were significantly correlated with FEV₁ and LCI.
PREFUL MRI is feasible in pediatric CF, distinguishes patients undergoing pulmonary exacerbations compared to healthy subjects, and correlates with HP ¹²⁹Xe MRI as well as functional measures of disease severity. PREFUL MRI does not require breath-holds and is straight forward to implement on any MRI scanner.

View all citing articles on Scopus

: This study was funded by the National Institutes of Health/National Institutes of Health Heart, Lung and Blood Institute (NHLBI) R01HL105643 with additional support from GE Healthcare (Wauwatosa, WI, United States). Analysis was conducted with additional support from the Duke Center for In Vivo Microscopy, the National Institutes of Health/National Institute of Biomedical Imaging and Bioengineering National Biomedical Technology Resource Center (P41 EB015897).

View full text

Original InvestigationExtending Semiautomatic Ventilation Defect Analysis for Hyperpolarized 129Xe Ventilation MRI

Rationale and Objectives

Materials and Methods

Results

Conclusions

Section snippets

Materials and methods

Linear-Binning Histogram Scaling

Advantages of the Corrected Linear-Binning Method

Conclusions

Acknowledgments

Acad Radiol

Acad Radiol

Pattern Recogn

Med Image Anal

Neuroimage

Hyperpolarized He and Xe MRI: differences in asthma before bronchodilation

J Magn Reson Imaging

Probing the regional distribution of pulmonary gas-exchange through single-breath, gas- and dissolved-phase 129Xe MR imaging

J Appl Physiol

Regional mapping of gas uptake by blood and tissue in the human lung using hyperpolarized xenon-129 MRI

J Magn Reson Imaging

Hyperpolarized Xe MR imaging of alveolar gas uptake in humans

PLoS One

DOE begins rationing helium-3

Phys Today

Quantitative analysis of hyperpolarized (129) Xe ventilation imaging in healthy volunteers and subjects with chronic obstructive pulmonary disease

NMR Biomed

Original Investigation
Extending Semiautomatic Ventilation Defect Analysis for Hyperpolarized ¹²⁹Xe Ventilation MRI