Correlating exhaled aerosol images to small airway obstructive diseases: A study with dynamic mode decomposition and machine learning

RESEARCH ARTICLE Correlating exhaled aerosol images to small airway obstructive diseases: A study with dynamic mode decomposition and machine learning Jinxiang Xi ID1,2*, Weizhong Zhao3 1 Department of Biomedical Engineering, California Baptist University, Riverside, California, United States of America, 2 Department of Aerospace and Mechanical Engineering, California Baptist University, Riverside, California, United States of America, 3 Division of Bioinformatics and Biostatistics, National Center for Toxicological Research, Jefferson, Arkansas, United States of America a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 * Abstract Background OPEN ACCESS Citation: Xi J, Zhao W (2019) Correlating exhaled aerosol images to small airway obstructive diseases: A study with dynamic mode decomposition and machine learning. PLoS ONE 14(1): e0211413. https://doi.org/10.1371/journal. pone.0211413 Editor: Roi Gurka, Coastal Carolina University, UNITED STATES Received: July 12, 2018 Accepted: January 14, 2019 Published: January 31, 2019 Copyright: © 2019 Xi, Zhao. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: The authors received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. Exhaled aerosols from lungs have unique patterns, and their variation can be correlated to the underlying lung structure and associated abnormities. However, it is challenging to characterize such aerosol patterns and differentiate their difference because of their complexity. This challenge is even greater for small airway diseases, where the disturbance signals are weak. Objectives and methods The objective of this study is exploiting different feature extraction algorithms to develop a practical classifier to diagnose obstructive lung diseases using exhaled aerosol images. These include proper orthogonal decomposition (POD), principal component analysis (PCA), dynamic mode decomposition (DMD), and DMD with control (DMDC). Aerosol images were generated via physiology-based simulations in one normal and four diseased airway models in G7-9 bronchioles. The image data were classified using both the support vector machine (SVM) and random forest (RF) algorithms. The effectiveness of different features was evaluated by classification accuracy and misclassification rate. Findings Results show a significantly higher performance using dynamic feature extractions (DMD and DMDC) than static algorithms (POD and PCA). Adding the control variables to DMD further improved classification accuracy. Comparing the classification methods, RF persistently outperformed SVM for all types of features considered. While the performance of RF constantly increased with the number of features retained, the performance of SVM peaked at 50 and decreased thereafter. The 5-class classification accuracy was 94.8% using the PLOS ONE | https://doi.org/10.1371/journal.pone.0211413 January 31, 2019 1 / 22 Correlating exhaled aerosol images to lung anomaly DMDC-RF model and 93.0% using the DMD-RF model, both of which were higher than 87.0% in the previous study that used fractal dimension features. Conclusion Considering that disease progression is inherently a dynamic process, DMD(C)-based feature extraction preserves temporal information and is preferred over POD and PCA. Compared with hand-crafted features like fractals, feature extraction by DMD and DMDC is automatic and more accurate. Introduction Lung diseases, either being restrictive (inhalation) such as acute respiratory distress syndrome (ARDS) and cystic fibrosis, or obstructive (exhalation) such as asthma and chronic obstructive pulmonary disease (COPD), will affect the respiratory airflow and cause a disturbance to the exhaled airflow pattern [1–3]. Exhaled aerosols can reveal a wealth of information about the health of the lungs [4]. However, there are many challenges to correlate these images to the underlying lung structural remodeling. The distributions of the exhaled aerosols are exceedingly complex, which are determined by the airflow and aerosol dynamics. Exhaled aerosol images from deep lungs generally cannot be differentiated by mere inspection. As a result, how to extract useful features from these seemingly chaotic observables is crucial in developing an effective algorithm to diagnose lung abnormalities based on exhaled aerosol images. In our previous studies [5–9], fractal-based features, such as lacunarity, fractal dimension (FD), and multifractal spectrum, have been explored for the quantification of aerosol images and subsequent machine learning of disease status. In combination with the random forest (RF) algorithm [10, 11], the optimal accuracy was predicted at 87.0% for a five-class classification of asthmatic diseases located in small airways (G8 bronchiole) [12]. Our hypothesis in this study is as follows. Instead of using FD (global or local) of the image which may suffer information loss [13, 14], aerosol patterns formed by exhaled airflows, together with their temporal dynamic processes, should better capture the progression of airway structural remodeling in deep lungs. Boser et al. [13] demonstrated that global FD could not accurately describe the asthmatic lungs and local features of the diseased region should be included. To avoid possible information loss, it is suggested that features, or eigenmodes, be extracted directly from the images (i.e., pixel values). By projecting the aerosol images onto low-dimensional eigenmodes, the underlying physics (fluid-particle transport equations) can be approximated by a dynamical system with fewer degrees of freedom, which can be used for the detection, monitoring, and when combined with targeted pulmonary drug delivery, treatment of the lung diseases. Great advances were made in extracting features or eigenmodes from numerical simulations and experimental visualizations. Proper orthogonal decomposition (POD) [15], principal component analysis (PCA) [16], global eigenmodes [17], balanced modes [18, 19], and dynamic mode decomposition (DMD) [20, 21] have given useful insights on the dynamics of fluid flows. POD decomposes the dynamics into orthogonal modes. It provides a low-rank basis and a hierarchy of features that are most predominant in the system. PCA is equivalent to POD but removes the mean to increase the contrast. In machine learning and pattern recognition PCA has been widely applied for modal decomposition and dimensionality reduction. In recent years, DMD has attracted attention in various fields as an approach for the above PLOS ONE | https://doi.org/10.1371/journal.pone.0211413 January 31, 2019 2 / 22 Co (...truncated)