Ching-Long Lin

Inhaled particles deposit non-uniformly in the extrathoracic airway, and where they concentrate determines the local tissue dose. Algebraic correlations predict only the total deposition efficiency, whereas computational fluid–particle dynamics (CFPD) resolves the spatial deposition field but must be repeated for every inhalation flow rate and particle size. We present graph neural network surrogates that predict the extrathoracic airway wall-deposition field from the wall-element (mesh) graph and two operating parameters—flow rate and particle diameter—without requiring flow-field input. The network is trained separately for four mouth-throat geometries: two idealized geometries [the United States Pharmacopeia (USP) and Virginia Commonwealth University (VCU)] and two subject-specific geometries (one male and one female) reconstructed from computed tomography (CT) images. Both total-deposition and element-wise losses are implemented to capture complementary global and local deposition characteristics, respectively. For the two subject-specific airways, the model reproduced the deposition field, including the concentrated laryngeal deposition core, and predicted the total deposition efficiency with coefficients of determination (R2) of 0.97–0.98. For the idealized geometries, the model recovered the core but not the surrounding low-dose region, where the CFPD reference was dominated by sampling noise. Once trained, the model predicts the deposition field at a new combination of flow rate and particle diameter in 3.3 ms, compared with approximately 3,000 s for the corresponding CFPD simulation. This acceleration reduces the time required for parametric sweeps from weeks to seconds, enabling practical optimization, uncertainty quantification, and performance assessment of inhalers and induction ports.

Objectives: To develop a contrastive learning model for lung disease classification using discriminative CT imaging embeddings. Methods: A total of 1,187 subjects were included: asthma ( n = 315), COPD ( n = 355), post-COVID-19 ( n = 375), and healthy controls ( n = 142). Of these, 1,003 subjects had a single visit with similarly protocoled CT scans acquired at total lung capacity (TLC) and residual volume (RV), and 92 (33 asthma and 59 post-COVID-19) completed a follow-up visit, with two scans per visit. We developed a modified contrastive learning model incorporating an expert-conditioned routing network and adaptive temperature scaling to learn discriminative embeddings. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC). The embeddings were further validated via k-means clustering, and quantitative CT (qCT) metrics were compared across the derived clusters. The embedding space was used to track disease progression or improvement in the follow-up disease subgroup and to evaluate the model's ability to predict qCT metrics, quantified by the coefficient of determination ( R 2 ) . Results: The model achieved a macro-AUC of 89.3% (95% CI: 86.5, 91.8; P < 0.001) in differentiating the four classes. Post-COVID-19 emerged as a distinct class from asthma and COPD in the t-SNE embedding space, and its embeddings across two visits captured disease improvement. Additionally, the learned embeddings showed predictive power for several qCT metrics, particularly the Jacobian ( R 2 =0.61). Conclusions: The proposed model effectively differentiated these three lung diseases and provided meaningful embeddings for phenotype characterization, longitudinal assessment, and qCT metric prediction.

The long-term consequences of COVID-19 remain poorly understood. To assess CT image-based biomarkers and clinical symptoms in COVID-19 survivors approximately 3-4 years after infection. Eighty post-COVID-19 participants (81% infected with the pre-Alpha strain) underwent pulmonary function tests (PFTs) and inspiratory/expiratory CT at approximately 5 months (Visit 1, V1) and 3-4 years (Visit 2, V2) after infection. At V2, participants completed the St. George's Respiratory Questionnaire (SGRQ), Leicester Cough Questionnaire (LCQ), Fatigue Severity Scale (FSS), modified Medical Research Council Dyspnea Scale (mMRC), and a study-specific symptom and medical history questionnaire. Seventy-eight healthy individuals served as controls. Image-based biomarkers included airway diameter, airway wall thickness, functional small airway disease percentage (fSAD%), ground-glass opacity percentage (GGO%), and bronchovascular percentage (Bronchovascular%). Post-COVID-19 participants exhibited normal predicted PFT values, but consistently lower DLCO compared with healthy controls at both visits. At V2, they also reported significantly worse SGRQ scores than the general population, indicating reduced quality of life. Although the elevated fSAD% and GGO% observed at V1 largely resolved by V2, several biomarkers of airway and vascular remodeling persisted, including increased Bronchovascular%, airway narrowing and wall thickening, and a compositional shift from large to small airways and a shift from small to large vessels. Persistent symptoms-such as fatigue, brain fog, cough, and hypertension-were associated with these structural abnormalities. Airway and vascular structural abnormalities persisted in COVID-19 survivors 3-4 years after infection and were associated with ongoing symptoms and reduced quality of life.

Accurate airway segmentation is essential for quantitative assessment of pulmonary diseases but remains challenging in expiratory computed tomography (CT) because airways become thinner and less distinguishable from adjacent vessels, resulting in severe class imbalance and increased peripheral leakages. Conventional manual or rule-based methods are time-consuming and limited reproducibility, while most deep learning studies have focused exclusively on inspiratory scans. To address these limitations, we propose a three-dimensional (3D) segmentation framework with an attention gate, termed Averaged Multi-Gaussian Response (AMGR), integrated into a U-Net architecture and tailored for expiratory airways. The AMGR gate stabilizes feature fusion by averaging multiple Gaussian responses per channel, suppressing peripheral noise and improving continuity in distal airway. The model is trained using a composite loss that combines cross-entropy, intersection over union, and centerline, jointly enhancing overlap accuracy and structural completeness. A total of 120 subjects from four cohorts (asthma, COPD, post-COVID-19, and healthy subjects) were used for training, validation, and independent testing. Quantitative evaluation demonstrated a high Dice score of 0.9592 relative to existing approaches, while maintaining balanced Precision (0.9416) and Recall (0.9780). Qualitatively, the model effectively suppresses false-positive leakages and noise. These results confirm that the AMGR framework provides a robust and generalizable solution for airway segmentation in expiratory CT, enabling accurate geometric analysis for future phenotype- and biomarker-based pulmonary studies.

Introduction: The aim of this research is to elucidate chronic obstructive pulmonary disease (COPD) progression by quantifying lung ventilation heterogeneities using single-photon emission computed tomography (SPECT) images and establishing correlations with quantitative computed tomography (qCT) imaging-based metrics. This approach seeks to enhance our understanding of how structural and functional changes influence ventilation heterogeneity in COPD.Methods: Eight COPD subjects completed a longitudinal study with three visits, spaced about a year apart. CT scans were performed at each visit and qCT-based variables were derived to measure the structural and functional characteristics of the lungs, while the SPECT-based variables were used to quantify lung ventilation heterogeneity. The correlations between key qCT-based variables and SPECT-based variables were examined.Results: The SPECT-based ventilation heterogeneity (CV_Total) showed strong correlations with the qCT-based functional small airway disease percentage (fSAD%_Total) and emphysematous tissue percentage (Emph%_Total) in the total lung, based on cross-sectional data. Over the 2-year period, changes in SPECT-based hot spots (TC_Max) exhibited strong negative correlations with changes in fSAD%_Total, Emph%_Total, and the average airway diameter in the left upper lobe, as well as a strong positive correlation with alternations in airflow distribution between the upper and lower lobes.Discussion: In conclusion, this study found strong positive cross-sectional correlations between CV_Total and both fSAD% and Emph%, suggesting that these markers primarily reflect static disease severity at a single time point. In contrast, longitudinal correlations between changes in TC_Max and other variables over 2 years may capture the dynamic process of hot spot formation, independent of disease severity. These findings suggest that changes in TC_Max may serve as a more sensitive biomarker than changes in CV_Total for tracking the underlying mechanisms of COPD progression.

The simulation of gas flow in the bronchial tree using computational fluid dynamics (CFD) has become a useful tool for the analysis of gas flow mechanics, structural deformation, ventilation, and particle deposition for drug delivery during spontaneous and assisted breathing. CFD allows for new hypotheses to be tested in silico, and detailed results generated without performing expensive experimental procedures that could be potentially harmful to patients. Such computational techniques are also useful for analyzing structure–function relationships in healthy and diseased lungs, assessing regional ventilation at various time points over the course of clinical treatment, or elucidating the changes in airflow patterns over the life span. CFD has also allowed for the development and use of image-based (i.e., patient-specific) models of three-dimensional (3D) airway trees with realistic boundary conditions to achieve more meaningful and personalized data that may be useful for planning effective treatment protocols. This focused review will present a summary of the techniques used in generating realistic 3D airway tree models, the limitations of such models, and the methodologies used for CFD airflow simulation. We will discuss mathematical and image-based geometric models, as well as the various boundary conditions that may be imposed on these geometric models. The results from simulations utilizing mathematical and image-based geometric models of the airway tree will also be discussed in terms of similarities to actual gas flow in the human lung.

Background: The objective of this study is to understand chronic obstructive pulmonary disease (COPD) phenotypes and their progressions by quantifying heterogeneities of lung ventilation from the single photon emission computed tomography (SPECT) images and establishing associations with the quantitative computed tomography (qCT) imaging-based clusters and variables. Methods: Eight COPD patients completed a longitudinal study of three visits with intervals of about a year. CT scans of these subjects at residual volume, functional residual capacity, and total lung capacity were taken for all visits. The functional and structural qCT-based variables were derived, and the subjects were classified into the qCT-based clusters. In addition, the SPECT variables were derived to quantify the heterogeneity of lung ventilation. The correlations between the key qCT-based variables and SPECT-based variables were examined. Results: The SPECT-based coefficient of variation (CV _Total ), a measure of ventilation heterogeneity, showed strong correlations (|r| ≥ 0.7) with the qCT-based functional small airway disease percentage (fSAD% _Total ) and emphysematous tissue percentage (Emph% _Total ) in the total lung on cross-sectional data. As for the two-year changes, the SPECT-based maximum tracer concentration (TC _max ), a measure of hot spots, exhibited strong negative correlations with fSAD% _Total , Emph% _Total , average airway diameter in the left upper lobe, and airflow distribution in the middle and lower lobes. Conclusion: Small airway disease is highly associated with the heterogeneity of ventilation in COPD lungs. TC _max is a more sensitive functional biomarker for COPD progression than CV _Total . Besides fSAD% _Total and Emph% _Total , segmental airways narrowing and imbalanced ventilation between upper and lower lobes may contribute to the development of hot spots over time.