Large-scale investigation of deep learning approaches for ventilated lung segmentation using multi-nuclear hyperpolarized gas MRI

Optimized quantitative mapping of cardiopulmonary oscillations using hyperpolarized 129 Xe gas exchange MRI: Digital phantoms and clinical evaluation in CTEPH

PURPOSE

The interaction between 129 Xe atoms and pulmonary capillary red blood cells provides cardiogenic signal oscillations that display sensitivity to precapillary and postcapillary pulmonary hypertension. Recently, such oscillations have been spatially mapped, but little is known about optimal reconstruction or sensitivity to artifacts. In this study, we use digital phantom simulations to specifically optimize keyhole reconstruction for oscillation imaging. We then use this optimized method to re-establish healthy reference values and quantitatively evaluate microvascular flow changes in patients with chronic thromboembolic pulmonary hypertension (CTEPH) before and after pulmonary thromboendarterectomy (PTE).

METHODS

A six-zone digital lung phantom was designed to investigate the effects of radial views, key radius, and SNR. One-point Dixon 129 Xe gas exchange MRI images were acquired in a healthy cohort (n = 17) to generate a reference distribution and thresholds for mapping red blood cell oscillations. These thresholds were applied to 10 CTEPH participants, with 6 rescanned following PTE.

RESULTS

For undersampled acquisitions, a key radius of0.14 k max $$ 0.14{k}_{\mathrm{max}} $$ was found to optimally resolve oscillation defects while minimizing excessive heterogeneity. CTEPH participants at baseline showed higher oscillation defect + low (32 ± 14%) compared with healthy volunteers (18 ± 12%, p < 0.001). For those scanned both before and after PTE, oscillation defect + low decreased from 37 ± 13% to 23 ± 14% (p = 0.03).

CONCLUSIONS

Digital phantom simulations have informed an optimized keyhole reconstruction technique for gas exchange images acquired with standard 1-point Dixon parameters. Our proposed methodology enables more robust quantitative mapping of cardiogenic oscillations, potentially facilitating effective regional quantification of microvascular flow impairment in patients with pulmonary vascular diseases such as CTEPH.

Hyperpolarized Gas Imaging in Lung Diseases: Functional and Artificial Intelligence Perspective

Pathophysiologic changes in lung diseases are often accompanied by changes in ventilation and gas exchange. Comprehensive evaluation of lung function cannot be obtained through chest X-ray and computed tomography. Proton-based lung MRI is particularly challenging due to low proton density within the lung tissue. In this review, we discuss an emerging technology--hyperpolarized gas MRI with inhaled 129Xe, which provides functional and microstructural information and has the potential as a clinical tool for detecting the early stage and progression of certain lung diseases. We review the hyperpolarized 129Xe MRI studies in patients with a range of pulmonary diseases, including chronic obstructive pulmonary disease, asthma, cystic fibrosis, pulmonary hypertension, radiation-induced lung injury and interstitial lung disease, and the applications of artificial intelligence were reviewed as well.

129Xe MRI Ventilation Defects in Asthma: What is the Upper Limit of Normal and Minimal Clinically Important Difference?

RATIONALE AND OBJECTIVES

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.

MATERIALS AND METHODS

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.

RESULTS

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.

CONCLUSION

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.

Deep Learning Approaches for Quantifying Ventilation Defects in Hyperpolarized Gas Magnetic Resonance Imaging of the Lung: A Review

This paper provides an in-depth overview of Deep Neural Networks and their application in the segmentation and analysis of lung Magnetic Resonance Imaging (MRI) scans, specifically focusing on hyperpolarized gas MRI and the quantification of lung ventilation defects. An in-depth understanding of Deep Neural Networks is presented, laying the groundwork for the exploration of their use in hyperpolarized gas MRI and the quantification of lung ventilation defects. Five distinct studies are examined, each leveraging unique deep learning architectures and data augmentation techniques to optimize model performance. These studies encompass a range of approaches, including the use of 3D Convolutional Neural Networks, cascaded U-Net models, Generative Adversarial Networks, and nnU-net for hyperpolarized gas MRI segmentation. The findings highlight the potential of deep learning methods in the segmentation and analysis of lung MRI scans, emphasizing the need for consensus on lung ventilation segmentation methods.

Implementable Deep Learning for Multi-sequence Proton MRI Lung Segmentation: A Multi-center, Multi-vendor, and Multi-disease Study

BACKGROUND

Recently, deep learning via convolutional neural networks (CNNs) has largely superseded conventional methods for proton (1 H)-MRI lung segmentation. However, previous deep learning studies have utilized single-center data and limited acquisition parameters.

PURPOSE

Develop a generalizable CNN for lung segmentation in 1 H-MRI, robust to pathology, acquisition protocol, vendor, and center.

STUDY TYPE

Retrospective.

POPULATION

A total of 809 1 H-MRI scans from 258 participants with various pulmonary pathologies (median age (range): 57 (6-85); 42% females) and 31 healthy participants (median age (range): 34 (23-76); 34% females) that were split into training (593 scans (74%); 157 participants (55%)), testing (50 scans (6%); 50 participants (17%)) and external validation (164 scans (20%); 82 participants (28%)) sets.

FIELD STRENGTH/SEQUENCE

1.5-T and 3-T/3D spoiled-gradient recalled and ultrashort echo-time 1 H-MRI.

ASSESSMENT

2D and 3D CNNs, trained on single-center, multi-sequence data, and the conventional spatial fuzzy c-means (SFCM) method were compared to manually delineated expert segmentations. Each method was validated on external data originating from several centers. Dice similarity coefficient (DSC), average boundary Hausdorff distance (Average HD), and relative error (XOR) metrics to assess segmentation performance.

STATISTICAL TESTS

Kruskal-Wallis tests assessed significances of differences between acquisitions in the testing set. Friedman tests with post hoc multiple comparisons assessed differences between the 2D CNN, 3D CNN, and SFCM. Bland-Altman analyses assessed agreement with manually derived lung volumes. A P value of <0.05 was considered statistically significant.

RESULTS

The 3D CNN significantly outperformed its 2D analog and SFCM, yielding a median (range) DSC of 0.961 (0.880-0.987), Average HD of 1.63 mm (0.65-5.45) and XOR of 0.079 (0.025-0.240) on the testing set and a DSC of 0.973 (0.866-0.987), Average HD of 1.11 mm (0.47-8.13) and XOR of 0.054 (0.026-0.255) on external validation data.

DATA CONCLUSION

The 3D CNN generated accurate 1 H-MRI lung segmentations on a heterogenous dataset, demonstrating robustness to disease pathology, sequence, vendor, and center.

EVIDENCE LEVEL

4.

TECHNICAL EFFICACY

Stage 1.

The argument for utilising magnetic resonance imaging as a tool for monitoring lung structure and function in pediatric patients

INTRODUCTION

Although historically challenging to perform in the lung, technological advancements have made Magnetic Resonance Imaging (MRI) increasingly applicable for pediatric pulmonary imaging. Furthermore, a wide array of functional imaging techniques has become available that may be leveraged alongside structural imaging for increasingly sensitive biomarkers, or as outcome measures in the evaluation of novel therapies.

AREAS COVERED

In this review, recent technical advancements and modern methodologies for structural and functional lung MRI are described. These include ultrashort echo time (UTE) MRI, free-breathing contrast agent-free, functional lung MRI, and hyperpolarized gas MRI, amongst other techniques. Specific examples of the application of these methods in children are provided, principally drawn from recent research in asthma, bronchopulmonary dysplasia, and cystic fibrosis.

EXPERT OPINION

Pediatric lung MRI is rapidly growing, and is well poised for clinical utilization, as well as continued research into early disease detection, disease processes, and novel treatments. Structure/function complementarity makes MRI especially attractive as a tool for increased adoption in the evaluation of pediatric lung disease. Looking toward the future, novel technologies, such as low-field MRI and artificial intelligence, mitigate some of the traditional drawbacks of lung MRI and will aid in improving access to MRI in general, potentially spurring increased adoption and demand for pulmonary MRI in children.

A Dual-Channel Deep Learning Approach for Lung Cavity Estimation From Hyperpolarized Gas and Proton MRI

BACKGROUND

Hyperpolarized gas MRI can quantify regional lung ventilation via biomarkers, including the ventilation defect percentage (VDP). VDP is computed from segmentations derived from spatially co-registered functional hyperpolarized gas and structural proton (1 H)-MRI. Although acquired at similar lung inflation levels, they are frequently misaligned, requiring a lung cavity estimation (LCE). Recently, single-channel, mono-modal deep learning (DL)-based methods have shown promise for pulmonary image segmentation problems. Multichannel, multimodal approaches may outperform single-channel alternatives.

PURPOSE

We hypothesized that a DL-based dual-channel approach, leveraging both 1 H-MRI and Xenon-129-MRI (129 Xe-MRI), can generate LCEs more accurately than single-channel alternatives.

STUDY TYPE

Retrospective.

POPULATION

A total of 480 corresponding 1 H-MRI and 129 Xe-MRI scans from 26 healthy participants (median age [range]: 11 [8-71]; 50% females) and 289 patients with pulmonary pathologies (median age [range]: 47 [6-83]; 51% females) were split into training (422 scans [88%]; 257 participants [82%]) and testing (58 scans [12%]; 58 participants [18%]) sets.

FIELD STRENGTH/SEQUENCE

1.5-T, three-dimensional (3D) spoiled gradient-recalled 1 H-MRI and 3D steady-state free-precession 129 Xe-MRI.

ASSESSMENT

We developed a multimodal DL approach, integrating 129 Xe-MRI and 1 H-MRI, in a dual-channel convolutional neural network. We compared this approach to single-channel alternatives using manually edited LCEs as a benchmark. We further assessed a fully automatic DL-based framework to calculate VDPs and compared it to manually generated VDPs.

STATISTICAL TESTS

Friedman tests with post hoc Bonferroni correction for multiple comparisons compared single-channel and dual-channel DL approaches using Dice similarity coefficient (DSC), average boundary Hausdorff distance (average HD), and relative error (XOR) metrics. Bland-Altman analysis and paired t-tests compared manual and DL-generated VDPs. A P value < 0.05 was considered statistically significant.

RESULTS

The dual-channel approach significantly outperformed single-channel approaches, achieving a median (range) DSC, average HD, and XOR of 0.967 (0.867-0.978), 1.68 mm (37.0-0.778), and 0.066 (0.246-0.045), respectively. DL-generated VDPs were statistically indistinguishable from manually generated VDPs (P = 0.710).

DATA CONCLUSION

Our dual-channel approach generated LCEs, which could be integrated with ventilated lung segmentations to produce biomarkers such as the VDP without manual intervention.

EVIDENCE LEVEL

4.

TECHNICAL EFFICACY

Stage 1.