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

Nasal anatomy and sniffing in respiration and olfaction of wild and domestic animals

Animals have been widely utilized as surrogate models for humans in exposure testing, infectious disease experiments, and immunology studies. However, respiratory diseases affect both humans and animals. These disorders can spontaneously affect wild and domestic animals, impacting their quality and quantity of life. The origin of such responses can primarily be traced back to the pathogens deposited in the respiratory tract. There is a lack of understanding of the transport and deposition of respirable particulate matter (bio-aerosols or viruses) in either wild or domestic animals. Moreover, local dosimetry is more relevant than the total or regionally averaged doses in assessing exposure risks or therapeutic outcomes. An accurate prediction of the total and local dosimetry is the crucial first step to quantifying the dose-response relationship, which in turn necessitates detailed knowledge of animals' respiratory tract and flow/aerosol dynamics within it. In this review, we examined the nasal anatomy and physiology (i.e., structure-function relationship) of different animals, including the dog, rat, rabbit, deer, rhombus monkey, cat, and other domestic and wild animals. Special attention was paid to the similarities and differences in the vestibular, respiratory, and olfactory regions among different species. The ventilation airflow and behaviors of inhaled aerosols were described as pertinent to the animals' mechanisms for ventilation modulation and olfaction enhancement. In particular, sniffing, a breathing maneuver that animals often practice enhancing olfaction, was examined in detail in different animals. Animal models used in COVID-19 research were discussed. The advances and challenges of using numerical modeling in place of animal studies were discussed. The application of this technique in animals is relevant for bidirectional improvements in animal and human health.

A comparison of CFPD, compartment, and uniform distribution models for radiation dosimetry of radionuclides in the lung

Radioactive aerosols that arise from natural sources and nuclear accidents can be a long-term hazard to human health. Despite the heterogeneous particle deposition in the respiratory tract, uniform aerosol doses have long been assumed in respiratory radiation dosimetry predictions, such as in the compartment and uniform distribution models. It is unclear how these deposition patterns affect internal radiation doses, which are critical in the health assessment of radioactive hazards. This work seeks to quantify the radio-dosimetry sensitivity to initial deposition patterns by comparing computational and compartment/uniform models. A new approach was developed to implement the compartment model into voxel phantoms (e.g. VIP-man) for radiation dosimetry. The calculated radiation fluence, energy deposition density and organ doses were compared to those obtained from coupling computational fluid-particle dynamics (CFPD) with Monte Carlo radiation transport and to those obtained from uniform source distribution approximation. The results show that the source particle distribution within the respiratory system substantially influences the radiation dosimetry distribution. The compartment and uniform models underestimated aerosol deposition in the crania ridge, leading to lower doses in the trachea and surrounding organs. For 0.5 MeV gammas, the CFPD-Monte Carlo N-particle (MCNP) model predicted a tracheal dose twice that of the compartment model and four times the uniform model. For 1 MeV betas, the CFPD-MCNP-predicted tracheal dose is 2.6 times that of the compartment model and 14 times the uniform model. Compared to the compartment/uniform models, the CFPD approach predicted a 50% lower beta dose in the lung but higher beta doses in the heart (six times), liver (four times) and stomach (2.5 times). It is suggested that including compartments for the lung periphery and tracheal carina ridge may improve the dosimetry accuracy of compartment models.

Deciphering Exhaled Aerosol Fingerprints for Early Diagnosis and Personalized Therapeutics of Obstructive Respiratory Diseases in Small Airways

Respiratory diseases often show no apparent symptoms at their early stages and are usually diagnosed when permanent damages have been made to the lungs. A major site of lung pathogenesis is the small airways, which make it highly challenging to detect using current techniques due to the diseases’ location (inaccessibility to biopsy) and size (below normal CT/MRI resolution). In this review, we present a new method for lung disease detection and treatment in small airways based on exhaled aerosols, whose patterns are uniquely related to the health of the lungs. Proof-of-concept studies are first presented in idealized lung geometries. We subsequently describe the recent developments in feature extraction and classification of the exhaled aerosol images to establish the relationship between the images and the underlying airway remodeling. Different feature extraction algorithms (aerosol density, fractal dimension, principal mode analysis, and dynamic mode decomposition) and machine learning approaches (support vector machine, random forest, and convolutional neural network) are elaborated upon. Finally, future studies and frequent questions related to clinical applications of the proposed aerosol breath testing are discussed from the authors’ perspective. The proposed breath testing has clinical advantages over conventional approaches, such as easy-to-perform, non-invasive, providing real-time feedback, and is promising in detecting symptomless lung diseases at early stages.

SARS COV-2 virus-laden droplets coughed from deep lungs: Numerical quantification in a single-path whole respiratory tract geometry

When an infected person coughs, many virus-laden droplets will be exhaled out of the mouth. Droplets from deep lungs are especially infectious because the alveoli are the major sites of coronavirus replication. However, their exhalation fraction, size distribution, and exiting speeds are unclear. This study investigated the behavior and fate of respiratory droplets (0.1-4 μm) during coughs in a single-path respiratory tract model extending from terminal alveoli to mouth opening. An experimentally measured cough waveform was used to control the alveolar wall motions and the flow boundary conditions at lung branches from G2 to G18. The mouth opening was modeled after the image of a coughing subject captured using a high-speed camera. A well-tested k-ω turbulence model and Lagrangian particle tracking algorithm were applied to simulate cough flow evolutions and droplet dynamics under four cough depths, i.e., tidal volume ratio (TVR) = 0.13, 0.20. 0.32, and 0.42. The results show that 2-μm droplets have the highest exhalation fraction, regardless of cough depths. A nonlinear relationship exists between the droplet exhalation fraction and cough depth due to a complex deposition mechanism confounded by multiscale airway passages, multiregime flows, and drastic transient flow effects. The highest exhalation fraction is 1.6% at the normal cough depth (TVR = 0.32), with a mean exiting speed of 20 m/s. The finding that most exhaled droplets from deep lungs are 2 μm highlights the need for more effective facemasks in blocking 2-μm droplets and smaller both in infectious source control and self-protection from airborne virus-laden droplets.

Spectral graph theory efficiently characterizes ventilation heterogeneity in lung airway networks

This paper introduces a linear operator for the purposes of quantifying the spectral properties of transport within resistive trees, such as airflow in lung airway networks. The operator, which we call the Maury matrix, acts only on the terminal nodes of the tree and is equivalent to the adjacency matrix of a complete graph summarizing the relationships between all pairs of terminal nodes. We show that the eigenmodes of the Maury operator have a direct physical interpretation as the relaxation, or resistive, modes of the network. We apply these findings to both idealized and image-based models of ventilation in lung airway trees and show that the spectral properties of the Maury matrix characterize the flow asymmetry in these networks more concisely than the Laplacian modes, and that eigenvector centrality in the Maury spectrum is closely related to the phenomenon of ventilation heterogeneity caused by airway narrowing or obstruction. This method has applications in dimensionality reduction in simulations of lung mechanics, as well as for characterization of models of the airway tree derived from medical images.

Machine Learning Applied to Diagnosis of Human Diseases: A Systematic Review

Human healthcare is one of the most important topics for society. It tries to find the correct effective and robust disease detection as soon as possible to patients receipt the appropriate cares. Because this detection is often a difficult task, it becomes necessary medicine field searches support from other fields such as statistics and computer science. These disciplines are facing the challenge of exploring new techniques, going beyond the traditional ones. The large number of techniques that are emerging makes it necessary to provide a comprehensive overview that avoids very particular aspects. To this end, we propose a systematic review dealing with the Machine Learning applied to the diagnosis of human diseases. This review focuses on modern techniques related to the development of Machine Learning applied to diagnosis of human diseases in the medical field, in order to discover interesting patterns, making non-trivial predictions and useful in decision-making. In this way, this work can help researchers to discover and, if necessary, determine the applicability of the machine learning techniques in their particular specialties. We provide some examples of the algorithms used in medicine, analysing some trends that are focused on the goal searched, the algorithm used, and the area of applications. We detail the advantages and disadvantages of each technique to help choose the most appropriate in each real-life situation, as several authors have reported. The authors searched Scopus, Journal Citation Reports (JCR), Google Scholar, and MedLine databases from the last decades (from 1980s approximately) up to the present, with English language restrictions, for studies according to the objectives mentioned above. Based on a protocol for data extraction defined and evaluated by all authors using PRISMA methodology, 141 papers were included in this advanced review.

Ventilation Modulation and Nanoparticle Deposition in Respiratory and Olfactory Regions of Rabbit Nose

The rabbit nose's ability to filter out inhaled agents is directly related to its defense to infectious diseases. The knowledge of the rabbit nose anatomy is essential to appreciate its functions in ventilation regulation, aerosol filtration and olfaction. The objective of this study is to numerically simulate the inhalation and deposition of nanoparticles in a New Zealand white (NZW) rabbit nose model with an emphasis on the structure-function relation under normal and sniffing conditions. To simulate the sniffing scenario, the original nose model was modified to generate new models with enlarged nostrils or vestibules based on video images of a rabbit sniffing. Ventilations into the maxilloturbinate and olfactory region were quantified with varying nostril openings, and deposition rates of inhaled aerosols ranging from 0.5 nm to 1000 nm were characterized on the total, sub-regional and local basis. Results showed that particles which deposited in the olfactory region came from a specific area in the nostril. The spiral vestibule played an essential role in regulating flow resistance and flow partition into different parts of the nose. Increased olfactory doses were persistently predicted in models with expanded nostrils or vestibule. Particles in the range of 5-50 nm are more sensitive to the geometry variation than other nanoparticles. It was also observed that exhaled aerosols occupy only the central region of the nostril, which minimized the mixing with the aerosols close to the nostril wall, and potentially allowed the undisruptive sampling of odorants. The results of this study shed new light on the ventilation regulation and inhalation dosimetry in the rabbit nose, which can be further implemented to studies of infectious diseases and immunology in rabbits.

Radiation Dosimetry of Inhaled Radioactive Aerosols: CFPD and MCNP Transport Simulations of Radionuclides in the Lung

Despite extensive efforts in studying radioactive aerosols, including the transmission of radionuclides in different chemical matrices throughout the body, the internal organ-specific radiation dose due to inhaled radioactive aerosols has largely relied on experimental deposition data and simplified human phantoms. Computational fluid-particle dynamics (CFPD) has proven to be a reliable tool in characterizing aerosol transport in the upper airways, while Monte Carlo based radiation codes allow accurate simulation of radiation transport. The objective of this study is to numerically assess the radiation dosimetry due to particles decaying in the respiratory tract from environmental radioactive exposures by coupling CFPD with Monte Carlo N-Particle code, version 6 (MCNP6). A physiologically realistic mouth-lung model extending to the bifurcation generation G9 was used to simulate airflow and particle transport within the respiratory tract. Polydisperse aerosols with different distributions were considered, and deposition distribution of the inhaled aerosols on the internal airway walls was quantified. The deposition mapping of radioactive aerosols was then registered to the respiratory tract of an image-based whole-body adult male model (VIP-Man) to simulate radiation transport and energy deposition. Computer codes were developed for geometry visualization, spatial normalization, and source card definition in MCNP6. Spatial distributions of internal radiation dosimetry were compared for different radionuclides (¹³¹I, ^134,137Cs, ⁹⁰Sr-⁹⁰Y, ¹⁰³Ru and ^239,240Pu) in terms of the radiation fluence, energy deposition density, and dose per decay.