Kun Qing

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

To develop a breathhold acquisition for regional mapping of ventilation and the fractions of hyperpolarized xenon‐129 (Xe129) dissolved in tissue (lung parenchyma and plasma) and red blood cells (RBCs), and to perform an exploratory study to characterize data obtained in human subjects.

Assessment of lung function in asthma and COPD using hyperpolarized 129Xe chemical shift saturation recovery spectroscopy and dissolved-phase MRI.

Magnetic‐resonance spectroscopy and imaging using hyperpolarized xenon‐129 show great potential for evaluation of the most important function of the human lung ‐‐ gas exchange. In particular, chemical shift saturation recovery (CSSR) xenon‐129 spectroscopy provides important physiological information for the lung as a whole by characterizing the dynamic process of gas exchange, while dissolved‐phase (DP) xenon‐129 imaging captures the time‐averaged regional distribution of gas uptake by lung tissue and blood. Herein, we present recent advances in assessing lung function using CSSR spectroscopy and DP imaging in a total of 45 subjects (23 healthy, 13 chronic obstructive pulmonary disease (COPD) and 9 asthma). From CSSR acquisitions, the COPD subjects showed red blood cell to tissue–plasma (RBC‐to‐TP) ratios below the average for the healthy subjects (p < 0.001), but significantly higher septal wall thicknesses as compared with the healthy subjects (p < 0.005); the RBC‐to‐TP ratios for the asthmatic subjects fell outside two standard deviations (either higher or lower) from the mean of the healthy subjects, although there was no statistically significant difference for the average ratio of the study group as a whole. Similarly, from the 3D DP imaging acquisitions, we found that all the ratios (TP to gas phase (GP), RBC to GP, RBC to TP) measured in the COPD subjects were lower than those from the healthy subjects (p < 0.05 for all ratios), while these ratios in the asthmatic subjects differed considerably between subjects. Despite having been performed at different lung inflation levels, the RBC‐to‐TP ratios measured by CSSR and 3D DP imaging were fairly consistent with each other, with a mean difference of 0.037 (ratios from 3D DP imaging larger). In ten subjects the RBC‐to‐GP ratios obtained from the 3D DP imaging acquisitions were also highly correlated with their diffusing capacity of the lung for carbon monoxide per unit alveolar volume ratios measured by pulmonary function testing (R = 0.91). Copyright

Atlas‐based estimation of lung and lobar anatomy in proton MRI

To propose an accurate methodological framework for automatically segmenting pulmonary proton MRI based on an optimal consensus of a spatially normalized library of annotated lung atlases.

Rapid acquisition of helium-3 and proton three-dimensional image sets of the human lung in a single breath-hold using compressed sensing.

To develop and validate a method for acquiring helium‐3 (3He) and proton (1H) three‐dimensional (3D) image sets of the human lung with isotropic spatial resolution within a 10‐s breath‐hold by using compressed sensing (CS) acceleration, and to assess the fidelity of undersampled images compared with fully sampled images.

Regional anisotropy of airspace orientation in the lung as assessed with hyperpolarized helium-3 diffusion MRI

To evaluate regional anisotropy of lung‐airspace orientation by assessing the dependence of helium‐3 (3He) apparent diffusion coefficient (ADC) values on the direction of diffusion sensitization at two field strengths.

Regional investigation of lung function and microstructure parameters by localized 129Xe chemical shift saturation recovery and dissolved-phase imaging: A reproducibility study

To evaluate the reproducibility and regional variation of parameters obtained from localized 129Xe chemical shift saturation recovery (CSSR) MR spectroscopy in healthy volunteers and patients with chronic obstructive pulmonary disease (COPD) and to compare the results to 129Xe dissolved‐phase MR imaging.

Deep learning-based quantification of abdominal fat on magnetic resonance images

Obesity is increasingly prevalent and associated with increased risk of developing type 2 diabetes, cardiovascular diseases, and cancer. Magnetic resonance imaging (MRI) is an accurate method for determination of body fat volume and distribution. However, quantifying body fat from numerous MRI slices is tedious and time-consuming. Here we developed a deep learning-based method for measuring visceral and subcutaneous fat in the abdominal region of mice. Congenic mice only differ from C57BL/6 (B6) Apoe knockout (Apoe-/-) mice in chromosome 9 that is replaced by C3H/HeJ genome. Male congenic mice had lighter body weight than B6-Apoe-/- mice after being fed 14 weeks of Western diet. Axial and coronal T1-weighted sequencing at 1-mm-thickness and 1-mm-gap was acquired with a 7T Bruker ClinScan scanner. A deep learning approach was developed for segmenting visceral and subcutaneous fat based on the U-net architecture made publicly available through the open-source ANTsRNet library—a growing repository of well-known neural networks. The volumes of subcutaneous and visceral fat measured through our approach were highly comparable with those from manual measurements. The Dice score, root-mean-square error (RMSE), and correlation analysis demonstrated the similarity between two methods in quantifying visceral and subcutaneous fat. Analysis with the automated method showed significant reductions in volumes of visceral and subcutaneous fat but not non-fat tissues in congenic mice compared to B6 mice. These results demonstrate the accuracy of deep learning in quantification of abdominal fat and its significance in determining body weight.

Convolutional Neural Networks with Template-Based Data Augmentation for Functional Lung Image Quantification

RATIONALE AND OBJECTIVES We propose an automated segmentation pipeline based on deep learning for proton lung MRI segmentation and ventilation-based quantification which improves on our previously reported methodologies in terms of computational efficiency while demonstrating accuracy and robustness. The large data requirement for the proposed framework is made possible by a novel template-based data augmentation strategy. Supporting this work is the open-source ANTsRNet-a growing repository of well-known deep learning architectures first introduced here. MATERIALS AND METHODS Deep convolutional neural network (CNN) models were constructed and trained using a custom multilabel Dice metric loss function and a novel template-based data augmentation strategy. Training (including template generation and data augmentation) employed 205 proton MR images and 73 functional lung MRI. Evaluation was performed using data sets of size 63 and 40 images, respectively. RESULTS Accuracy for CNN-based proton lung MRI segmentation (in terms of Dice overlap) was left lung: 0.93 ± 0.03, right lung: 0.94 ± 0.02, and whole lung: 0.94 ± 0.02. Although slightly less accurate than our previously reported joint label fusion approach (left lung: 0.95 ± 0.02, right lung: 0.96 ± 0.01, and whole lung: 0.96 ± 0.01), processing time is <1 second per subject for the proposed approach versus ∼30 minutes per subject using joint label fusion. Accuracy for quantifying ventilation defects was determined based on a consensus labeling where average accuracy (Dice multilabel overlap of ventilation defect regions plus normal region) was 0.94 for the CNN method; 0.92 for our previously reported method; and 0.90, 0.92, and 0.94 for expert readers. CONCLUSION The proposed framework yields accurate automated quantification in near real time. CNNs drastically reduce processing time after offline model construction and demonstrate significant future potential for facilitating quantitative analysis of functional lung MRI.

Probing Changes in Lung Physiology in COPD Using CT, Perfusion MRI, and Hyperpolarized Xenon-129 MRI

RATIONALE AND OBJECTIVES Chronic obstructive pulmonary disease (COPD) is highly heterogeneous and not well understood. Hyperpolarized xenon-129 (Xe129) magnetic resonance imaging (MRI) provides a unique way to assess important lung functions such as gas uptake. In this pilot study, we exploited multiple imaging modalities, including computed tomography (CT), gadolinium-enhanced perfusion MRI, and Xe129 MRI, to perform a detailed investigation of changes in lung morphology and functions in COPD. Utility and strengths of Xe129 MRI in assessing COPD were also evaluated against the other imaging modalities. MATERIALS AND METHODS Four COPD patients and four age-matched normal subjects participated in this study. Lung tissue density measured by CT, perfusion measures from gadolinium-enhanced MRI, and ventilation and gas uptake measures from Xe129 MRI were calculated for individual lung lobes to assess regional changes in lung morphology and function, and to investigate correlations among the different imaging modalities. RESULTS No significant differences were found for all measures among the five lobes in either the COPD or age-matched normal group. Strong correlations (R > 0.5 or < -0.5, p < 0.001) were found between ventilation and perfusion measures. Also gas uptake by blood as measured by Xe129 MRI showed strong correlations with CT tissue density and ventilation measures (R > 0.5 or < -0.5, p < 0.001) and moderate to strong correlations with perfusion measures (R > 0.4 or < -0.5, p < 0.01). Four distinctive patterns of functional abnormalities were found in patients with COPD. CONCLUSION Xe129 MRI has high potential to uniquely identify multiple changes in lung physiology in COPD using a single breath-hold acquisition.

Signal‐to‐noise ratio, T2, and T2* for hyperpolarized helium‐3 MRI of the human lung at three magnetic field strengths

To evaluate T2, T2* , and signal‐to‐noise ratio (SNR) for hyperpolarized helium‐3 (3He) MRI of the human lung at three magnetic field strengths ranging from 0.43T to 1.5T.