Accurate, real-time tracking of airborne viruses is crucial for understanding transmission dynamics and mitigating the spread of respiratory illnesses like COVID-19. Current technologies, such as real-time polymerase chain reactions (RT-PCR), lack the necessary in situ and real-time capabilities for physicochemical characterization of airborne viruses. These methods typically require sample collection and processing, hindering the ability to observe the dynamic behavior and transformations of virus-laden particles in their natural environment. This limitation significantly impacts our understanding of viral transmission pathways, which include inhalation of viral droplets and aerosols, deposition onto mucous membranes, and contact with contaminated surfaces. Factors like viral load, duration of contact, environmental conditions, and host immunity further complicate transmission dynamics. Although studies confirm the importance of airborne transmission, particularly evidenced by the effectiveness of face masks in reducing SARS-CoV-2 transmission, in-situ, real-time observation methods are lacking. The study addresses this critical knowledge gap by developing a new technology to provide the needed in-situ and real-time capability for physicochemical tracking of airborne viruses.

Existing methods for observing, collecting, and quantifying viruses include molecular assays, immunoassays, multistage collectors, fluorescent sensors, and various types of polymerase chain reactions. While these techniques offer certain advantages, they are not in situ or real-time, limiting their ability to understand the dynamic changes that viruses undergo in air. Bioaerosols, including virus-laden droplets, are complex systems that undergo physicochemical transformations through interactions with gases and other airborne particles, influenced by factors such as temperature, humidity, and air dynamics. The lack of real-time observation hinders our understanding of these vital processes, including viral aerosol evaporation and condensation, which are fundamental to primary and secondary viral transmission.

This research employed Nano-DIHM, a lensless holographic microscopy technique, coupled with a gas flow tube. This setup allowed real-time observation of single or ensembles of viral particles as they flowed through the imaging volume. Nano-DIHM directly records interference patterns (holograms) of scattered light, and numerical reconstruction using Octopus/Stingray software recovers object information, including size, phase, shape, and surface properties. The study utilized both active MS2 bacteriophage (a common viral surrogate) and inactivated SARS-CoV-2 particles. The MS2 was purchased from ZeptoMetrix, while inactivated SARS-CoV-2 was provided by McGill University's Contaminant Level 3 Platform and confirmed via whole-genome sequencing. Experiments involved observing viruses in both air and water, under dynamic (simulating coughing and sneezing) and static conditions. High-resolution scanning/transmission electron microscopy (S/TEM) and Talos-S/TEM validated Nano-DIHM results, and particle size analyzers (PSA, SMPS, OPS) provided additional size distribution data. Artificial intelligence (AI), implemented within the Stingray software, enabled 4D tracking and automated object classification. Experiments included analyzing MS2 and SARS-CoV-2 particles in mixtures with various organic (alpha-pinene, olive oil, honey) and inorganic (TiO2, Fe2O3) compounds to assess the system's ability to distinguish viruses in complex matrices. The Stingray software was trained using over 10,000 holograms and 100,000 iterations, achieving approximately 99% accuracy in SARS-CoV-2 identification.

The Nano-DIHM system successfully validated MS2 bacteriophage size and morphology measurements against S/TEM and Talos-S/TEM data, demonstrating its accuracy in determining particle size, shape, and surface properties. The size distribution of aerosolized MS2 particles (dry aerosols) ranged from nano- to microscale, consistent with SMPS and OPS measurements. The phase and intensity analysis revealed morphological variations, including spherical and irregular shapes, potentially due to particle coagulation or aggregation during aerosolization. 4D tracking of MS2 particles demonstrated their dynamic trajectories in air, showing morphology evolution over time, likely related to particle interactions with water or self-assembly. Experiments with mixtures of MS2 and TiO2 indicated physicochemical transformations. The study demonstrated in-situ, real-time detection and classification of SARS-CoV-2 particles in both dynamic (simulating sneezing and coughing at different velocities) and static modes, confirming the morphology found in previous high-resolution microscopy studies. The 3D size distribution of SARS-CoV-2 droplets varied from ~300 nm to several micrometers, correlating with findings from previous studies on coughing, sneezing, and breathing. AI-powered Stingray software enabled automated detection and classification of SARS-CoV-2 with high accuracy in mixed samples (approximately 99%), differentiating it from MS2 and other components. Further experiments demonstrated the capability of Nano-DIHM to distinguish MS2 particles with various coatings (TiO2, olive oil, alpha-pinene, honey), showing that these coatings significantly impacted particle morphology and size. Finally, the surface properties (edge gradient and roughness) of SARS-CoV-2, MS2, and metal oxide particles were distinct, indicating the potential of Nano-DIHM to characterize surface features.

The successful detection, classification, and physicochemical characterization of airborne viruses, particularly SARS-CoV-2, using Nano-DIHM represents a significant advancement in real-time virus tracking. The ability to observe dynamic changes in particle size, shape, and morphology, coupled with automated analysis, provides crucial insights into viral transmission mechanisms. The findings address the critical need for in situ, real-time technologies to study viral aerosol dynamics. The results on the effects of different coatings highlight the impact of environmental factors on viral aerosol behavior, informing more comprehensive modeling of viral transmission. The high accuracy of the AI-powered classification demonstrates the potential for rapid and reliable virus detection, with implications for public health responses.

This study demonstrates the successful development and validation of Nano-DIHM as a rapid, real-time, in situ method for detecting, classifying, and characterizing airborne viruses. The high accuracy of automated detection using AI, coupled with the ability to observe dynamic morphological changes, offers significant potential for improving understanding of virus transmission and informing public health interventions. Future research should focus on expanding the library of classifiers to include a wider range of viruses and environmental pollutants. Further development could lead to portable, point-of-care diagnostic tools for rapid detection of respiratory viruses.

The study used inactivated SARS-CoV-2, and the behavior of live viruses may differ. The accuracy of the AI-based classification might decrease with increasingly complex sample matrices, necessitating further library expansion. Although background holograms were recorded to minimize contamination, the possibility of some impurities affecting the results cannot be completely excluded. The current system may produce false positives, although ongoing improvements in classifiers aim to address this.