EGElias Graf

Elias Graf

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Elias Graf has a total of 84 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Evaluating automated pollen monitoring for thunderstorm asthma and public health forecasting: a comparative study in Melbourne, Australia

    by Edwin R. Lampugnani, Usha Nattala, Danny Csutoros, Kale Sniderman, Beau Picking, Kathryn M. Emmerson, Elias Graf, Yanick Zeder, Erny Niederberger, Ed Newbigin

    High-quality pollen measurements are essential for effective environmental monitoring and reliable public health forecasting. Automated pollen counters offer continuous, high-resolution observations that have the potential to reduce reliance on labour-intensive manual microscopy, but their performance must be evaluated against established methods. This study evaluates the performance of the SwisensPoleno Mars by comparing its daily and sub-daily pollen measurements of total pollen, and the two dominant airborne pollen types in Melbourne, grass and cypress, with those from a co-located Burkard Hirst-type spore trap at the University of Melbourne over the 92-day spring monitoring period in 2022. Data from the automated system showed a strong linear relationship with the co-located manual trap, with coefficients of determination ( R 2 ) of 0.92 for total daily pollen counts, 0.92 for grass pollen, and 0.86 for cypress pollen. Grass pollen measurements, which are central to epidemic thunderstorm asthma (ETSA) risk forecasting in Victoria, showed particularly high correspondence, including consistent classification into low, moderate, and high operational thresholds. The SwisensPoleno Mars also reproduced expected diurnal variation in pollen concentrations, demonstrating its ability to capture short-term changes in airborne pollen levels. These results indicate that automated pollen counters can deliver reliable observations at both daily and sub-daily scales while substantially reducing labour demands. The strong performance of the automated system supports its use for routine pollen monitoring, its suitability for supplementing or replacing manual counting in Victoria and its value and suitability for applications that benefit from high temporal resolution, including ETSA risk forecasting.

  • Open Access

    Advancing automated identification of airborne fungal spores: guidelines for cultivation and reference dataset creation

    by Nicolas Bruffaerts, Elias Graf, Predrag Matavulj, Astha Tiwari, Ioanna Pyrri, Yanick Zeder, Sophie Erb, Maria Plaza, Silas Dietler, Tommaso Bendinelli, et al.

    Airborne bioparticles, including fungal spores, are of major concern for human and plant health, necessitating precise monitoring systems. While a European norm exists for manual volumetric monitoring, there's a growing interest in automated real-time methods. However, these methods rely heavily on machine learning, facing challenges due to diverse particle characteristics and limited training data availability, especially for fungal spores. This study aims to address this gap by outlining best practices for collecting reference material and creating tailored datasets for training algorithms. Using 17 fungal species from the Belgian fungi collection BCCM/IHEM, including five Alternaria species, key aspects such as in vitro cultivation, dry spore harvest, and aerosolization were addressed. Simple classification models were developed, achieving varying accuracies on different monitors. The Plair Rapid-E+ demonstrated accuracies ranging from 83.4% to 95.1% (macro average F1-score 0.61), with better recognition for Cladosporium spp. and Curvularia caricae-papayae. The SwisensPoleno Jupiter, initially achieving a macro average F1-score of 0.77 with holographic images of eight genera, improved to 0.83 when combined with fluorescence data. Accuracies ranged from 55 to 95%, with notable performance for Alternaria spp. and Curvularia caricae-papayae. Species differentiation was also shown to be possible for Cladosporium, but was more difficult for some Alternaria species, while the macro average F1-score remained good (0.72). Overall, this protocol paves the way for more efficient, standard, and accurate automatic identification of airborne fungal spores.


  • Open Access

    Method for calibration of bioaerosol monitors based on reference pollen aerosols and imaging-based Lagrangian particle tracking as reference: the case of the Swisens Poleno

    by Stefan Horender, Christina Giannakoudaki, Reto Abt, Kevin Auderset, Benoît Crouzy, Sophie Erb, Oguzhan Erdogdu, Elias Graf, Kenjiro Iida, Erny Niederberger, et al.

    Several automated pollen monitors have recently become available, most using conventional and/or machine-learning algorithms to detect pollen and classify their taxa. An international intercomparison campaign was organised by the EUMETNET AutoPollen Programme and the ADOPT COST Action in Munich (March–July 2021) (Maya-Manzano et al. 2023). The study showed that some automatic systems, especially those that have built-in correction factors to compensate for losses of sampling, detection and classification, agreed with the averaged pollen concentration of four manual Hirst-type measurements. The manual pollen counting method, relying on human operators, is still used today due to a lack of traceable metrological standards for pollen monitoring. Additionally, existing reference particle counters have been validated up to particle sizes around 20 µm only (Vasilatou et al. 2022). In this study, we evaluated the shake-the-box particle tracking/detection algorithm (Novara et al. 2023) to serve as a reference measurement of particle number concentration. A volume is illuminated with an expanded light sheet produced by a double-pulse laser, and four calibrated double-frame cameras record images of the particles inside this volume. The algorithm determines the position and velocity of each particle, allowing the measurement of particle number concentration directly in the air without sampling inlets that could influence the measurement.Particle tracking was validated against the Inkjet Aerosol Generator at the Japanese National Metrology Institute NMIJ and the reference optical particle counter at METAS using an 11-D spectrometer (Grimm GmbH, Germany) as a transfer standard and size-certified polystyrene particles. Expanded uncertainties (coverage factor k=2) were 24 %, 10 %, and 7 % for particle sizes 15 μm, 20 μm, and 26 μm, respectively. The good agreement among the three methods shows that the shake-the-box particle tracking method can be applied as a reference for particle number concentration.We used a laboratory-based method for characterising the performance of bioaerosol monitors as a whole unit (hardware plus identification algorithms) using the particle tracking method and, in a separate experiment, freshly sampled pollen. Experiments were carried out with the SwisensPoleno Jupiter, which combines light scattering, inline digital holography and ultraviolet laser-induced fluorescence with machine learning to classify different particles. For a pollen grain to be measured, it must be sampled, detected and correctly classified. Each stage is subject to particle losses, leading to a measurement efficiency below 100 %. Particle tracking measurements of pollen (Pinus, Zea Mays) resulted in an average counting efficiency of about 43 %, neglecting particle losses in the Sigma-2 sampling head and issues with the classification of particles. The unit-to-unit variability of the Poleno instruments was 30 % based on measurement with three units. Independent experiments with Alnus glutinosa, Betula pendula and Corylus avellana showed that the major source of losses, however, originates from the pollen classification algorithm, which is trimmed to best correlate with Hirst measurements (currently the defacto reference for pollen monitoring). This highlights the need for an independent, standardised method for evaluating classifier losses.

  • Open Access

    The influence of local particles on classifier performance for pollen monitoring

    by Andreas Schwendimann, Yanick Zeder, Kilian Koch, Elias Graf, Erny Niederberger, Elena Gottardini, Antonella Cristofori, Fabiana Cristofolini, Magdalena Widmann

    Pollen detection through automatic instruments has significantly improved over the recent years. Hirst-type traps start to be complemented with automatic instruments throughout Europe. Instead of the traditional identification of airborne pollen particles using light microscopy, data of the particles is collected and subsequently analysed by an AI-classifier. As an airflow-cytometer, SwisensPoleno instruments capture a fingerprint of each particle in-flight. Holography images, as well as fluorescence spectral data, as for SwisensPoleno Jupiter, are collected to classify measured pollen grains flowing through the device in real-time. Labelled datasets of each pollen-type of interest are required for the training of a new classifier. These are generated by collecting fresh and pure pollen from plants and aerosolizing it directly into a SwisensPoleno instrument under controlled conditions. New classifiers are evaluated by correlating concentrations determined by the automatic instrument with daily concentrations of co-located Hirst-type traps. To evaluate the performance of classifiers throughout Europe, multiple sites in different countries are assessed. Using results of the latest pollen classifier “Swisens (2025)”, based on holography images and fluorescence spectra, we show here how different locations of the SwisensPoleno and different input data can affect the resulting correlations to Hirst data, as well as the Mean Absolute Error (MAE). Differences in performance are expected between sites which are far apart, due to many factors such as differing geography, local climatic conditions and flora. Bad performance may arise from unknown interfering particles, only abundant at one specific site, and thus not included in the training datasets.Our results demonstrate that this effect can also occur at a sub-regional scale, in sites only 30 km apart (Figure 1A), installed with the same instrument, and analysing data with the same classifier. Fraxinus pollen concentrations for P48 (Bolzano, Italy) correlate very well with a co-located Hirst (Figure 1B); that isn’t true for P46 (San Michele all’Adige, Italy), where strong interference from other particles is present during late May and beginning of June (Figure 1C). Interfering particles, similar to the target taxa, require site specific fine-tuning, e.g. in form of additional filters specifically excluding the interfering particles. In conclusion, while automatic pollen detection instruments show great promise in improving accuracy and efficiency, our findings highlight the importance of site-specific adaptations to address geographic and environmental variability, ensuring reliable performance across diverse locations.

  • Open Access

    Testing the generation of fungal spore aerosols with a new atomization setup

    by Elias Graf, Haruna Gütlin, Erny Niederberger, Philipp Burch, Tomke Musa

    To study bioaerosols under controlled conditions, aerosol chambers equipped with aerosol generators have been used for a long time. However, the method used for generation can change the constitution and properties of the bioaerosol produced, including the biological integrity of fungal spores, bacteria or airborne viruses. The properties of a bioaerosol in turn influence the results of detection, enumeration and identification methods downstream (Pogner et al., 2024). Recent developments of automatic bioaerosol monitors equipped with AI-based identification algorithms require simple and reliable generation of bioaerosols in the laboratory to collect data for the machine learning trainings.Figure 1 View into the SAG chamber through the window at the front, containing a petri dish with a fungal colony.The Swisens Aerosol Generator (SAG) is an atomizer for efficient and gentle aerosolization of fungal spores, as well as other dry biological materials, for measurement with the SwisensPoleno and other instruments. The SAG principle is based on the design described by Lee et al. (2010), with improvements for better controllability and a higher yield. It consists of a chamber, shown in Figure 2, in which the petri dishes and other materials are placed. Pressured air, generated by a separate air supply unit, is led to a nozzle inside the chamber and directed over the biological materials. The horizontal air stream detaches the fungal spores into the air of the chamber, which can then be taken in by the instrument to measure the content. Figure 2 Left: Aerosol chamber with HEPA filter at the top for clean air supply. Right: Air supply unit with digital flow meter, flow controls and air filter.The structure and quantity of aerosolized particles is highly dependent on the fungi species, its growth stage and success, as well as the airflow onto the petri dish. Measurements with Cladosporium cladosporioides (Tested by Pogner et al. (2024) with SAG prototype) created, for example, numerous data of agglomerated spores and mycelium parts besides single spores, whereas Alternaria alternata constantly generated single spores in high numbers. With the SAG the process and the efficiency of fungal spores aerosolization can be improved, making it a new tool besides other fungal spore aerosols generation methods. This opens more opportunities to choose the best means of aerosolization depending on the fungal species and required particles. References:Lee, Jun Hyun, Gi Byung Hwang, Jae Hee Jung, Dae Hee Lee, und Byung Uk Lee. 2010. «Generation characteristics of fungal spore and fragment bioaerosols by airflow control over fungal cultures». Journal of Aerosol Science 41 (3): 319–25. https://doi.org/10.1016/j.jaerosci.2009.11.002.Pogner, Clara-E, Elias Graf, Erny Niederberger, und Markus Gorfer. 2024. «What do spore particles look like - use of real-time measurements and holography imaging to view spore particles from four bioaerosol generators». Aerosol Science and Technology 58 (7): 1–17. https://doi.org/10.1080/02786826.2024.2338544.

  • Open Access

    Merging holography, fluorescence, and machine learning for in situ continuous characterization and classification of airborne microplastics

    by Nicholas D. Beres, Julia Burkart, Elias Graf, Yanick Zeder, Lea Ann Dailey, Bernadett Weinzierl

    The continued increase in global plastic production and poor waste management ensures that plastic pollution will be a serious environmental concern for years to come. Because of their size, shape, and relatively low density, plastic particles between 1 and 1000 µm in size (known as microplastics or MPs) emitted directly into the environment (“primary”) or created due to degradation (“secondary”) may be transported through the atmosphere, similarly to other coarse-mode particles such as mineral dust. MPs can thus be advected over great distances, reaching even the most pristine and remote areas of Earth, and may have significant negative consequences for humans and the environment. The detection and analysis of MPs once airborne, however, remains a challenge because most observational methods are offline and resource-intensive and, therefore, not capable of providing continuous quantitative information. In this study, we present results using an online in situ airflow cytometer (SwisensPoleno Jupiter; Swisens AG; Emmen, Switzerland) – coupled with machine learning – to detect, analyze, and classify airborne single-particle MPs in near real time. The performance of the instrument in differentiating between single-particle MPs of five common polymer types (including polypropylene, polyethylene, polyamide, poly(methyl methacrylate), and polyethylene terephthalate) was investigated under laboratory conditions using combined information about their size and shape (determined using holographic imaging) and fluorescence measured using three excitation wavelengths and five emission detection windows. The classification capability using these methods was determined alongside other coarse-mode aerosol particles with similar morphology or fluorescence characteristics, such as a mineral dust and several pollen taxa. The tested MPs exhibit a measurable fluorescence signal that not only allows them to be distinguished from other fluorescent particles, such as pollen, but also differentiated from each other, with high (> 90 %) classification accuracy based on their multispectral fluorescence signatures. The classification accuracies of machine learning models using only holographic images of particles, only the fluorescence response, and combined information from holography and fluorescence to predict particle types are presented and compared. The last model, using both the holographic images and fluorescence information for each particle, was the most optimal model used, providing the highest classification accuracy compared to employing models using only the holography or fluorescence response separately. The results provide a foundation for significantly improving the understanding of the properties and types of MPs present in the atmosphere.


  • Open Access

    Reply on RC1

    by Nicholas D. Beres, Julia Burkart, Elias Graf, Yanick Zeder, Lea Ann Dailey, Bernadett Weinzierl

    The continued increase in global plastic production and poor waste management ensures that plastic pollution is a serious environmental concern for years to come. Because of their size, shape, and relatively low density, plastic particles between 1–1000 μm in size (known as microplastics, or MPs) emitted directly into the environment (“primary”) or created due to degradation (“secondary”) may be transported through the atmosphere, similar to other coarse-mode particles, such as mineral dust. MPs can thus be advected over great distances, reaching even the most pristine and remote areas of the Earth, and may have significant negative consequences for humans and the environment. The detection and analysis of MPs once airborne, however, remains a challenge because most observational methods are offline and resource-intensive, and, therefore, are not capable of providing continuous quantitative information. In this study, we present results using an online, in situ airflow cytometer (SwisensPoleno Jupiter; Swisens AG; Emmen, Switzerland) – coupled with machine learning – to detect, analyze, and classify airborne, single-particle MPs in near real time. The performance of the instrument to differentiate single-particle MPs of five common polymer types (including polypropylene, polyethylene, polyamide, poly(methyl methacrylate), and polyethylene terephthalate) was investigated under laboratory conditions using combined information about their size and shape (determined using holographic imaging) and fluorescence measured using three excitation wavelengths and five emission detection windows. The classification capability using these methods was determined alongside other coarse-mode aerosol particles with similar morphology or fluorescence characteristics, such as a mineral dust and several pollen taxa. The tested MPs exhibit a measurable fluorescence signal that not only allows them to be distinguished from the other fluorescent particles, such as pollen, but can also be differentiated from each other, with high (> 90 %) classification accuracy based on their multispectral fluorescence signatures. The classification accuracies of machine learning models using only holographic images of particles, only the fluorescence response, and combined information from holography and fluorescence to predict particle type are presented and compared. The results provide a foundation towards significantly improving the understanding of the properties and types of MPs present in the atmosphere.

  • Open Access

    Reply on RC2

    by Nicholas D. Beres, Julia Burkart, Elias Graf, Yanick Zeder, Lea Ann Dailey, Bernadett Weinzierl

    The continued increase in global plastic production and poor waste management ensures that plastic pollution is a serious environmental concern for years to come. Because of their size, shape, and relatively low density, plastic particles between 1–1000 μm in size (known as microplastics, or MPs) emitted directly into the environment (“primary”) or created due to degradation (“secondary”) may be transported through the atmosphere, similar to other coarse-mode particles, such as mineral dust. MPs can thus be advected over great distances, reaching even the most pristine and remote areas of the Earth, and may have significant negative consequences for humans and the environment. The detection and analysis of MPs once airborne, however, remains a challenge because most observational methods are offline and resource-intensive, and, therefore, are not capable of providing continuous quantitative information. In this study, we present results using an online, in situ airflow cytometer (SwisensPoleno Jupiter; Swisens AG; Emmen, Switzerland) – coupled with machine learning – to detect, analyze, and classify airborne, single-particle MPs in near real time. The performance of the instrument to differentiate single-particle MPs of five common polymer types (including polypropylene, polyethylene, polyamide, poly(methyl methacrylate), and polyethylene terephthalate) was investigated under laboratory conditions using combined information about their size and shape (determined using holographic imaging) and fluorescence measured using three excitation wavelengths and five emission detection windows. The classification capability using these methods was determined alongside other coarse-mode aerosol particles with similar morphology or fluorescence characteristics, such as a mineral dust and several pollen taxa. The tested MPs exhibit a measurable fluorescence signal that not only allows them to be distinguished from the other fluorescent particles, such as pollen, but can also be differentiated from each other, with high (> 90 %) classification accuracy based on their multispectral fluorescence signatures. The classification accuracies of machine learning models using only holographic images of particles, only the fluorescence response, and combined information from holography and fluorescence to predict particle type are presented and compared. The results provide a foundation towards significantly improving the understanding of the properties and types of MPs present in the atmosphere.

  • Open Access

    Reply on RC1

    by Nicholas D. Beres, Julia Burkart, Elias Graf, Yanick Zeder, Lea Ann Dailey, Bernadett Weinzierl

    The continued increase in global plastic production and poor waste management ensures that plastic pollution is a serious environmental concern for years to come. Because of their size, shape, and relatively low density, plastic particles between 1–1000 μm in size (known as microplastics, or MPs) emitted directly into the environment (“primary”) or created due to degradation (“secondary”) may be transported through the atmosphere, similar to other coarse-mode particles, such as mineral dust. MPs can thus be advected over great distances, reaching even the most pristine and remote areas of the Earth, and may have significant negative consequences for humans and the environment. The detection and analysis of MPs once airborne, however, remains a challenge because most observational methods are offline and resource-intensive, and, therefore, are not capable of providing continuous quantitative information. In this study, we present results using an online, in situ airflow cytometer (SwisensPoleno Jupiter; Swisens AG; Emmen, Switzerland) – coupled with machine learning – to detect, analyze, and classify airborne, single-particle MPs in near real time. The performance of the instrument to differentiate single-particle MPs of five common polymer types (including polypropylene, polyethylene, polyamide, poly(methyl methacrylate), and polyethylene terephthalate) was investigated under laboratory conditions using combined information about their size and shape (determined using holographic imaging) and fluorescence measured using three excitation wavelengths and five emission detection windows. The classification capability using these methods was determined alongside other coarse-mode aerosol particles with similar morphology or fluorescence characteristics, such as a mineral dust and several pollen taxa. The tested MPs exhibit a measurable fluorescence signal that not only allows them to be distinguished from the other fluorescent particles, such as pollen, but can also be differentiated from each other, with high (> 90 %) classification accuracy based on their multispectral fluorescence signatures. The classification accuracies of machine learning models using only holographic images of particles, only the fluorescence response, and combined information from holography and fluorescence to predict particle type are presented and compared. The results provide a foundation towards significantly improving the understanding of the properties and types of MPs present in the atmosphere.

  • Open Access

    Comment on egusphere-2023-2853

    by Nicholas D. Beres, Julia Burkart, Elias Graf, Yanick Zeder, Lea Ann Dailey, Bernadett Weinzierl

    The continued increase in global plastic production and poor waste management ensures that plastic pollution is a serious environmental concern for years to come. Because of their size, shape, and relatively low density, plastic particles between 1–1000 μm in size (known as microplastics, or MPs) emitted directly into the environment (“primary”) or created due to degradation (“secondary”) may be transported through the atmosphere, similar to other coarse-mode particles, such as mineral dust. MPs can thus be advected over great distances, reaching even the most pristine and remote areas of the Earth, and may have significant negative consequences for humans and the environment. The detection and analysis of MPs once airborne, however, remains a challenge because most observational methods are offline and resource-intensive, and, therefore, are not capable of providing continuous quantitative information. In this study, we present results using an online, in situ airflow cytometer (SwisensPoleno Jupiter; Swisens AG; Emmen, Switzerland) – coupled with machine learning – to detect, analyze, and classify airborne, single-particle MPs in near real time. The performance of the instrument to differentiate single-particle MPs of five common polymer types (including polypropylene, polyethylene, polyamide, poly(methyl methacrylate), and polyethylene terephthalate) was investigated under laboratory conditions using combined information about their size and shape (determined using holographic imaging) and fluorescence measured using three excitation wavelengths and five emission detection windows. The classification capability using these methods was determined alongside other coarse-mode aerosol particles with similar morphology or fluorescence characteristics, such as a mineral dust and several pollen taxa. The tested MPs exhibit a measurable fluorescence signal that not only allows them to be distinguished from the other fluorescent particles, such as pollen, but can also be differentiated from each other, with high (> 90 %) classification accuracy based on their multispectral fluorescence signatures. The classification accuracies of machine learning models using only holographic images of particles, only the fluorescence response, and combined information from holography and fluorescence to predict particle type are presented and compared. The results provide a foundation towards significantly improving the understanding of the properties and types of MPs present in the atmosphere.