Deep neural network analyses of spirometry for structural phenotyping of chronic obstructive pulmonary disease

Curve-Modelling and Machine Learning for a Better COPD Diagnosis

Background

Development of new tools in artificial intelligence has an outstanding performance in the recognition of multidimensional patterns, which is why they have proven to be useful in the diagnosis of Chronic Obstructive Pulmonary Disease (COPD).

Methods

This was an observational analytical single-centre study in patients with spirometry performed in outpatient medical care. The segment that goes from the peak expiratory flow to the forced vital capacity was modelled with quadratic polynomials, the coefficients obtained were used to train and test neural networks in the task of classifying patients with COPD.

Results

A total of 695 patient records were included in the analysis. The COPD group was significantly older than the No COPD group. The pre-bronchodilator (Pre BD) and post-bronchodilator (Post BD) spirometric curves were modelled with a quadratic polynomial, and the coefficients obtained were used to feed three neural networks (Pre BD, Post BD and all coefficients). The best neural network was the one that used the post-bronchodilator coefficients, which has an input layer of 3 neurons and three hidden layers with sigmoid activation function and two neurons in the output layer with softmax activation function. This system had an accuracy of 92.9% accuracy, a sensitivity of 88.2% and a specificity of 94.3% when assessed using expert judgment as the reference test. It also showed better performance than the current gold standard, especially in specificity and negative predictive value.

Conclusion

Artificial Neural Networks fed with coefficients obtained from quadratic and cubic polynomials have interesting potential of emulating the clinical diagnostic process and can become an important aid in primary care to help diagnose COPD in an early stage.

Deep learning parametric response mapping from inspiratory chest CT scans: a new approach for small airway disease screening

OBJECTIVES

Parametric response mapping (PRM) enables the evaluation of small airway disease (SAD) at the voxel level, but requires both inspiratory and expiratory chest CT scans. We hypothesize that deep learning PRM from inspiratory chest CT scans can effectively evaluate SAD in individuals with normal spirometry.

METHODS

We included 537 participants with normal spirometry, a history of smoking or secondhand smoke exposure, and divided them into training, tuning, and test sets. A cascaded generative adversarial network generated expiratory CT from inspiratory CT, followed by a UNet-like network predicting PRM using real inspiratory CT and generated expiratory CT. The performance of the prediction is evaluated using SSIM, RMSE and dice coefficients. Pearson correlation evaluated the correlation between predicted and ground truth PRM. ROC curves evaluated predicted PRMfSAD (the volume percentage of functional small airway disease, fSAD) performance in stratifying SAD.

RESULTS

Our method can generate expiratory CT of good quality (SSIM 0.86, RMSE 80.13 HU). The predicted PRM dice coefficients for normal lung, emphysema, and fSAD regions are 0.85, 0.63, and 0.51, respectively. The volume percentages of emphysema and fSAD showed good correlation between predicted and ground truth PRM (|r| were 0.97 and 0.64, respectively, p < 0.05). Predicted PRMfSAD showed good SAD stratification performance with ground truth PRMfSAD at thresholds of 15%, 20% and 25% (AUCs were 0.84, 0.78, and 0.84, respectively, p < 0.001).

CONCLUSION

Our deep learning method generates high-quality PRM using inspiratory chest CT and effectively stratifies SAD in individuals with normal spirometry.

The Challenges of Spirometric Diagnosis of COPD

Chronic obstructive pulmonary disease (COPD) is one of the top causes of morbidity and mortality worldwide. Although for many years its accurate diagnosis has been a focus of intense research, it is still challenging. Due to its simplicity, portability, and low cost, spirometry has been established as the main tool to detect this condition, but its flawed performance makes it an imperfect COPD diagnosis gold standard. This review aims to provide an up-to-date literature overview of recent studies regarding COPD diagnosis; we seek to identify their limitations and establish perspectives for spirometric diagnosis of COPD in the XXI century by combining deep clinical knowledge of the disease with advanced computer analysis techniques.

Deep Learning Utilizing Suboptimal Spirometry Data to Improve Lung Function and Mortality Prediction in the UK Biobank

Background

Spirometry measures lung function by selecting the best of multiple efforts meeting pre-specified quality control (QC), and reporting two key metrics: forced expiratory volume in 1 second (FEV1) and forced vital capacity (FVC). We hypothesize that discarded submaximal and QC-failing data meaningfully contribute to the prediction of airflow obstruction and all-cause mortality.

Methods

We evaluated volume-time spirometry data from the UK Biobank. We identified "best" spirometry efforts as those passing QC with the maximum FVC. "Discarded" efforts were either submaximal or failed QC. To create a combined representation of lung function we implemented a contrastive learning approach, Spirogram-based Contrastive Learning Framework (Spiro-CLF), which utilized all recorded volume-time curves per participant and applied different transformations (e.g. flow-volume, flow-time). In a held-out 20% testing subset we applied the Spiro-CLF representation of a participant's overall lung function to 1) binary predictions of FEV1/FVC < 0.7 and FEV1 Percent Predicted (FEV1PP) < 80%, indicative of airflow obstruction, and 2) Cox regression for all-cause mortality.

Findings

We included 940,705 volume-time curves from 352,684 UK Biobank participants with 2-3 spirometry efforts per individual (66.7% with 3 efforts) and at least one QC-passing spirometry effort. Of all spirometry efforts, 24.1% failed QC and 37.5% were submaximal. Spiro-CLF prediction of FEV1/FVC < 0.7 utilizing discarded spirometry efforts had an Area under the Receiver Operating Characteristics (AUROC) of 0.981 (0.863 for FEV1PP prediction). Incorporating discarded spirometry efforts in all-cause mortality prediction was associated with a concordance index (c-index) of 0.654, which exceeded the c-indices from FEV1 (0.590), FVC (0.559), or FEV1/FVC (0.599) from each participant's single best effort.

Interpretation

A contrastive learning model using raw spirometry curves can accurately predict lung function using submaximal and QC-failing efforts. This model also has superior prediction of all-cause mortality compared to standard lung function measurements.

Funding

MHC is supported by NIH R01HL137927, R01HL135142, HL147148, and HL089856.BDH is supported by NIH K08HL136928, U01 HL089856, and an Alpha-1 Foundation Research Grant.DH is supported by NIH 2T32HL007427-41EKS is supported by NIH R01 HL152728, R01 HL147148, U01 HL089856, R01 HL133135, P01 HL132825, and P01 HL114501.PJC is supported by NIH R01HL124233 and R01HL147326.SPB is supported by NIH R01HL151421 and UH3HL155806.TY, FH, and CYM are employees of Google LLC.

Development and Internal Validation of Risk Assessment Models for Chronic Obstructive Pulmonary Disease in Coal Workers

Coal workers are more likely to develop chronic obstructive pulmonary disease due to exposure to occupational hazards such as dust. In this study, a risk scoring system is constructed according to the optimal model to provide feasible suggestions for the prevention of chronic obstructive pulmonary disease in coal workers. Using 3955 coal workers who participated in occupational health check-ups at Gequan mine and Dongpang mine of Hebei Jizhong Energy from July 2018 to August 2018 as the study subjects, random forest, logistic regression, and convolutional neural network models are established, and model performance is evaluated to select the optimal model, and finally a risk scoring system is constructed according to the optimal model to achieve model visualization. The training set results show that the logistic, random forest, and CNN models have sensitivities of 78.55%, 86.89%, and 77.18%; specificities of 85.23%, 92.32%, and 87.61%; accuracies of 81.21%, 85.40%, and 83.02%; Brier scores of 0.14, 0.10, and 0.14; and AUCs of 0.76, 0.88, and 0.78, respectively, and similar results are obtained for the test set and validation set, with the random forest model outperforming the other two models. The risk scoring system constructed according to the importance ranking of random forest predictor variables has an AUC of 0.842; the evaluation results of the risk scoring system shows that its accuracy rate is 83.7% and the AUC is 0.827, and the established risk scoring system has good discriminatory ability. The random forest model outperforms the CNN and logistic regression models. The chronic obstructive pulmonary disease risk scoring system constructed based on the random forest model has good discriminatory power.

Deep Learning-based Approach to Predict Pulmonary Function at Chest CT

Background Low-dose chest CT screening is recommended for smokers with the potential for lung function abnormality, but its role in predicting lung function remains unclear. Purpose To develop a deep learning algorithm to predict pulmonary function with low-dose CT images in participants using health screening services. Materials and Methods In this retrospective study, participants underwent health screening with same-day low-dose CT and pulmonary function testing with spirometry at a university affiliated tertiary referral general hospital between January 2015 and December 2018. The data set was split into a development set (model training, validation, and internal test sets) and temporally independent test set according to first visit year. A convolutional neural network was trained to predict the forced expiratory volume in the first second of expiration (FEV₁) and forced vital capacity (FVC) from low-dose CT. The mean absolute error and concordance correlation coefficient (CCC) were used to evaluate agreement between spirometry as the reference standard and deep-learning prediction as the index test. FVC and FEV₁ percent predicted (hereafter, FVC% and FEV₁%) values less than 80% and percent of FVC exhaled in first second (hereafter, FEV₁/FVC) less than 70% were used to classify participants at high risk. Results A total of 16 148 participants were included (mean age, 55 years ± 10 [SD]; 10 981 men) and divided into a development set (n = 13 428) and temporally independent test set (n = 2720). In the temporally independent test set, the mean absolute error and CCC were 0.22 L and 0.94, respectively, for FVC and 0.22 L and 0.91 for FEV₁. For the prediction of the respiratory high-risk group, FVC%, FEV₁%, and FEV₁/FVC had respective accuracies of 89.6% (2436 of 2720 participants; 95% CI: 88.4, 90.7), 85.9% (2337 of 2720 participants; 95% CI: 84.6, 87.2), and 90.2% (2453 of 2720 participants; 95% CI: 89.1, 91.3) in the same testing data set. The sensitivities were 61.6% (242 of 393 participants; 95% CI: 59.7, 63.4), 46.9% (226 of 482 participants; 95% CI: 45.0, 48.8), and 36.1% (91 of 252 participants; 95% CI: 34.3, 37.9), respectively. Conclusion A deep learning model applied to volumetric chest CT predicted pulmonary function with relatively good performance. © RSNA, 2023 Supplemental material is available for this article.

Deep learning using multilayer perception improves the diagnostic acumen of spirometry: a single-centre Canadian study

RATIONALE

Spirometry and plethysmography are the gold standard pulmonary function tests (PFT) for diagnosis and management of lung disease. Due to the inaccessibility of plethysmography, spirometry is often used alone but this leads to missed or misdiagnoses as spirometry cannot identify restrictive disease without plethysmography. We aimed to develop a deep learning model to improve interpretation of spirometry alone.

METHODS

We built a multilayer perceptron model using full PFTs from 748 patients, interpreted according to international guidelines. Inputs included spirometry (forced vital capacity, forced expiratory volume in 1 s, forced mid-expiratory flow_25-75), plethysmography (total lung capacity, residual volume) and biometrics (sex, age, height). The model was developed with 2582 PFTs from 477 patients, randomly divided into training (80%), validation (10%) and test (10%) sets, and refined using 1245 previously unseen PFTs from 271 patients, split 50/50 as validation (136 patients) and test (135 patients) sets. Only one test per patient was used for each of 10 experiments conducted for each input combination. The final model was compared with interpretation of 82 spirometry tests by 6 trained pulmonologists and a decision tree.

RESULTS

Accuracies from the first 477 patients were similar when inputs included biometrics+spirometry+plethysmography (95%±3%) vs biometrics+spirometry (90%±2%). Model refinement with the next 271 patients improved accuracies with biometrics+pirometry (95%±2%) but no change for biometrics+spirometry+plethysmography (95%±2%). The final model significantly outperformed (94.67%±2.63%, p<0.01 for both) interpretation of 82 spirometry tests by the decision tree (75.61%±0.00%) and pulmonologists (66.67%±14.63%).

CONCLUSIONS

Deep learning improves the diagnostic acumen of spirometry and classifies lung physiology better than pulmonologists with accuracies comparable to full PFTs.

Fx-Net and PureNet: Convolutional Neural Network architecture for discrimination of Chronic Obstructive Pulmonary Disease from smokers and healthy subjects through electronic nose signals

As one of the most reliable and significant indicators, Chronic Obstructive Pulmonary Disease (COPD) becomes a robust predictor of lung cancer early detection, the world's leading cause of cancer death. One of the methods is to analyze the Volatile Organic Compounds (VOCs) in exhaled breath using electronic noses (E-noses), which have become emerging tools for analyzing breath because of their potential and promising technology for diagnosing. However, the signal processing of the E-Nose sensor becomes vital in exposing information about the subject condition, which most researchers strive to accomplish. We proposed a Convolutional Neural Network (CNN) architecture to classify COPD in smokers and non-smokers, healthy subjects, and smokers from E-Nose signals to contribute to this field. Two models were constructed following E-Nose signal processing state-of-the-arts. One was by combined feature extraction and classifier, and the second was by CNN, which directly processed the raw signal. In addition, various feature extraction and classifier (Machine Learning and CNN) used in prior research were investigated. Using 3K and 5K Fold cross-validation results demonstrated that our proposed models outperformed in Kernel Principal Component Analysis (KPCA) with Fx-ConvNet and Pure-ConvNet. They all reached maximum F1-Score with zero standard deviation values indicating a consistent result. Further experiments also showed that KPCA contributed to the increasing performance of some classifiers with average F1-Score 0.933 and 0.068 as standard deviation values.

Assessment of Multi-Layer Perceptron Neural Network for Pulmonary Function Test’s Diagnosis Using ATS and ERS Respiratory Standard Parameters

The aim of the research work is to investigate the operability of the entire 23 pulmonary function parameters, which are stipulated by the American Thoracic Society (ATS) and the European Respiratory Society (ERS), to design a medical decision support system capable of classifying the pulmonary function tests into normal, obstructive, restrictive, or mixed cases. The 23 respiratory parameters specified by the ATS and the ERS guidelines, obtained from the Pulmonary Function Test (PFT) device, were employed as input features to a Multi-Layer Perceptron (MLP) neural network. Thirteen possible MLP Back Propagation (BP) algorithms were assessed. Three different categories of respiratory diseases were evaluated, namely obstructive, restrictive, and mixed conditions. The framework was applied on 201 PFT examinations: 103 normal and 98 abnormal cases. The PFT decision support system’s outcomes were compared with both the clinical truth (physician decision) and the PFT built-in diagnostic software. It yielded 92–99% and 87–92% accuracies on the training and the test sets, respectively. An 88–94% area under the receiver operating characteristic curve (ROC) was recorded on the test set. The system exceeded the performance of the PFT machine by 9%. All 23 ATS\ERS standard PFT parameters can be used as inputs to design a PFT decision support system, yielding a favorable performance compared with the literature and the PFT machine’s diagnosis program.

ERS/ATS technical standard on interpretive strategies for routine lung function tests

BACKGROUND

Appropriate interpretation of pulmonary function tests (PFTs) involves the classification of observed values as within/outside the normal range based on a reference population of healthy individuals, integrating knowledge of physiological determinants of test results into functional classifications and integrating patterns with other clinical data to estimate prognosis. In 2005, the American Thoracic Society (ATS) and European Respiratory Society (ERS) jointly adopted technical standards for the interpretation of PFTs. We aimed to update the 2005 recommendations and incorporate evidence from recent literature to establish new standards for PFT interpretation.

METHODS

This technical standards document was developed by an international joint Task Force, appointed by the ERS/ATS with multidisciplinary expertise in conducting and interpreting PFTs and developing international standards. A comprehensive literature review was conducted and published evidence was reviewed.

RESULTS

Recommendations for the choice of reference equations and limits of normal of the healthy population to identify individuals with unusually low or high results are discussed. Interpretation strategies for bronchodilator responsiveness testing, limits of natural changes over time and severity are also updated. Interpretation of measurements made by spirometry, lung volumes and gas transfer are described as they relate to underlying pathophysiology with updated classification protocols of common impairments.

CONCLUSIONS

Interpretation of PFTs must be complemented with clinical expertise and consideration of the inherent biological variability of the test and the uncertainty of the test result to ensure appropriate interpretation of an individual's lung function measurements.

Quantitative imaging analysis detects subtle airway abnormalities in symptomatic military deployers

Background

Exposure to inhalational hazards during post-9/11 deployment to Southwest Asia and Afghanistan puts military personnel at risk for respiratory symptoms and disease. Pulmonary function and qualitative chest high resolution computed tomography (HRCT) are often normal in “deployers” with persistent respiratory symptoms. We explored the utility of quantitative HRCT imaging markers of large and small airways abnormalities, including airway wall thickness, emphysema, and air trapping, in symptomatic deployers with clinically-confirmed lung disease compared to controls.

Methods

Chest HRCT images from 45 healthy controls and 82 symptomatic deployers with asthma, distal lung disease or both were analyzed using Thirona Lung quantification software to calculate airway wall thickness (by Pi10), emphysema (by percentage of lung volume with attenuation < -950 Hounsfield units [LAA%-950]), and three parameters of air trapping (expiratory/inspiratory total lung volume and mean lung density ratios, and LAA%-856). SAS v.9.4 was used to compare demographic and clinical characteristics between deployers and controls using Chi-Square, Fisher Exact or t-tests. Linear regression was used to assess relationships between pulmonary function and quantitative imaging findings.

Results

Gender and smoking status were not statistically significantly different between groups, but deployers were significantly younger than controls (42 vs 58 years, p < 0.0001), had higher body mass index (31 vs 28 kg/m², p = 0.01), and had fewer total smoking pack-years (8 vs. 26, p = 0.007). Spirometric measures were not statistically significantly different between groups. Pi10 and LAA%-950 were significantly elevated in deployers compared to controls in unadjusted analyses, with the emphysema measure remaining significantly higher in deployers after adjustment for age, sex, smoking, BMI, and expiratory total lung volume. Air trapping parameters were more common in control images, likely due to differences in age and smoking between groups. Among deployers, LAA%-950 and Pi10 were significantly correlated with spirometric markers of obstruction based on ratio of forced expiratory volume in one second (FEV1)/forced vital capacity (FVC) and/or percent predicted FEV1.

Conclusions

Quantitative chest HRCT imaging analysis identifies emphysema in deployers with asthma and distal lung disease, and may be useful in detecting and monitoring deployment-related lung disease in a population where spirometry is typically normal.

Treatment Trials in Young Patients with Chronic Obstructive Pulmonary Disease and Pre-Chronic Obstructive Pulmonary Disease Patients: Time to Move Forward

Chronic obstructive pulmonary disease (COPD) is the end result of a series of dynamic and cumulative gene-environment interactions over a lifetime. The evolving understanding of COPD biology provides novel opportunities for prevention, early diagnosis, and intervention. To advance these concepts, we propose therapeutic trials in two major groups of subjects: "young" individuals with COPD and those with pre-COPD. Given that lungs grow to about 20 years of age and begin to age at approximately 50 years, we consider "young" patients with COPD those patients in the age range of 20-50 years. Pre-COPD relates to individuals of any age who have respiratory symptoms with or without structural and/or functional abnormalities, in the absence of airflow limitation, and who may develop persistent airflow limitation over time. We exclude from the current discussion infants and adolescents because of their unique physiological context and COPD in older adults given their representation in prior randomized controlled trials (RCTs). We highlight the need of RCTs focused on COPD in young patients or pre-COPD to reduce disease progression, providing innovative approaches to identifying and engaging potential study subjects. We detail approaches to RCT design, including potential outcomes such as lung function, patient-reported outcomes, exacerbations, lung imaging, mortality, and composite endpoints. We critically review study design components such as statistical powering and analysis, duration of study treatment, and formats to trial structure, including platform, basket, and umbrella trials. We provide a call to action for treatment RCTs in 1) young adults with COPD and 2) those with pre-COPD at any age.

Analyzing the use of artificial intelligence for the management of chronic obstructive pulmonary disease (COPD)

OBJECTIVE

Chronic obstructive pulmonary disease (COPD) is a disease that causes airflow limitation to the lungs and has a high morbidity around the world. The objective of this study was to evaluate how artificial intelligence (AI) is being applied for the management of the disease, analyzing the objectives that are raised, the algorithms that are used and what results they offer.

METHODS

We conducted a scoping review following the Arksey and O'Malley (2005) and Levac et al. (2010) guidelines. Two reviewers independently searched, analyzed and extracted data from papers of five databases: Web of Science, PubMed, Scopus, Cinahl and Cochrane. To be included, the studies had to apply some AI techniques for the management of at least one stage of the COPD clinical process. In the event of any discrepancy between both reviewers, the criterion of a third reviewer prevailed.

RESULTS

380 papers were identified through database searches. After applying the exclusion criteria, 67 papers were included in the study. The studies were of a different nature and pursued a wide range of objectives, highlighting mainly those focused on the identification, classification and prevention of the disease. Neural nets, support vector machines and decision trees were the AI algorithms most commonly used. The mean and median values of all the performance metrics evaluated were between 80% and 90%.

CONCLUSIONS

The results obtained show a growing interest in the development of medical applications that manage the different phases of the COPD clinical process, especially predictive models. According to the performance shown, these models could be a useful complementary tool in the decision-making by health specialists, although more high-quality ML studies are needed to endorse the findings of this study.

[Pulmonary Emphysema: Visual Interpretation and Quantitative Analysis]

Pulmonary emphysema is a cause of chronic obstructive pulmonary disease. Emphysema can be accurately diagnosed via CT. The severity of emphysema can be assessed using visual interpretation or quantitative analysis. Various studies on emphysema using deep learning have also been conducted. Although the classification of emphysema has proven clinically useful, there is a need to improve the reliability of the measurement.

Application of Machine Learning in Pulmonary Function Assessment Where Are We Now and Where Are We Going?

Analysis of pulmonary function tests (PFTs) is an area where machine learning (ML) may benefit clinicians, researchers, and the patients. PFT measures spirometry, lung volumes, and carbon monoxide diffusion capacity of the lung (DLCO). The results are usually interpreted by the clinicians using discrete numeric data according to published guidelines. PFT interpretations by clinicians, however, are known to have inter-rater variability and the inaccuracy can impact patient care. This variability may be caused by unfamiliarity of the guidelines, lack of training, inadequate understanding of lung physiology, or simply mental lapses. A rules-based automated interpretation system can recapitulate expert’s pattern recognition capability and decrease errors. ML can also be used to analyze continuous data or the graphics, including the flow-volume loop, the DLCO and the nitrogen washout curves. These analyses can discover novel physiological biomarkers. In the era of wearables and telehealth, particularly with the COVID-19 pandemic restricting PFTs to be done in the clinical laboratories, ML can also be used to combine mobile spirometry results with an individual’s clinical profile to deliver precision medicine. There are, however, hurdles in the development and commercialization of the ML-assisted PFT interpretation programs, including the need for high quality representative data, the existence of different formats for data acquisition and sharing in PFT software by different vendors, and the need for collaboration amongst clinicians, biomedical engineers, and information technologists. Hurdles notwithstanding, the new developments would represent significant advances that could be the future of PFT, the oldest test still in use in clinical medicine.

Artificial Intelligence and Machine Learning in Chronic Airway Diseases: Focus on Asthma and Chronic Obstructive Pulmonary Disease

Chronic airway diseases are characterized by airway inflammation, obstruction, and remodeling and show high prevalence, especially in developing countries. Among them, asthma and chronic obstructive pulmonary disease (COPD) show the highest morbidity and socioeconomic burden worldwide. Although there are extensive guidelines for the prevention, early diagnosis, and rational treatment of these lifelong diseases, their value in precision medicine is very limited. Artificial intelligence (AI) and machine learning (ML) techniques have emerged as effective methods for mining and integrating large-scale, heterogeneous medical data for clinical practice, and several AI and ML methods have recently been applied to asthma and COPD. However, very few methods have significantly contributed to clinical practice. Here, we review four aspects of AI and ML implementation in asthma and COPD to summarize existing knowledge and indicate future steps required for the safe and effective application of AI and ML tools by clinicians.

Artificial Intelligence in COPD: New Venues to Study a Complex Disease

Chronic obstructive pulmonary disease (COPD) is a complex and heterogeneous disease that can benefit from novel approaches to understanding its evolution and divergent trajectories. Artificial intelligence (AI) has revolutionized how we can use clinical, imaging, and molecular data to understand and model complex systems. AI has shown impressive results in areas related to automated clinical decision making, radiological interpretation and prognostication. The unique nature of COPD and the accessibility to well-phenotyped populations result in an ideal scenario for AI development. This review provides an introduction to AI and deep learning and presents some recent successes in applying AI in COPD. Finally, we will discuss some of the opportunities, challenges, and limitations for AI applications in the context of COPD.