Automatic pollen recognition with the Rapid-E particle counter: the first-level procedure, experience and next steps

Abstract

Abstract. Pollen-induced allergy is among the most-prevalent non-contagious diseases, with about a quarter of European population sensitive to various atmospheric bioaerosols. In most European countries, pollen information is based on a weekly-cycle Hirst-type pollen trap method. This method is labour-intensive, requires narrow specialization abilities and substantial time, so that the pollen data are always delayed, subject to sampling- and counting-related uncertainties. Emerging new approaches to automatic pollen monitoring can, in principle, allow for real-time availability of the data with no human involvement. The goal of the current paper is to evaluate the capabilities of the new Plair Rapid-E pollen monitor and to construct the first-level pollen recognition algorithm. The evaluation was performed for three devices located in Lithuania, Serbia and Switzerland, with independent calibration data and classification algorithms. The Rapid-E output data include multi-angle scattering images and the fluorescence spectra recorded at several times for each particle reaching the device. Both modalities of the Rapid-E output were treated with artificial neural networks (ANN) and the results were combined to obtain the pollen type. For the first classification experiment, the monitor was challenged with a large variety of pollen types and the quality of many-to-many classification was evaluated. It was shown that in this case, both scattering- and fluorescence- based recognition algorithms fall short of acceptable quality. The combinations of these algorithms performed better exceeding 80 % accuracy for 5 out of 11 species. Fluorescence spectra showed similarities among different species ending up with three well-resolved groups: (Alnus, Corylus, Betula and Quercus), (Salix and Populus), and (Festuca, Artemisia, Juniperus). Within these groups, pollen is practically non-distinguishable for the first-level recognition procedure. Construction of multi-steps algorithms with sequential discrimination of pollen inside each group seems to be one of possible ways forwards. In order to connect the classification experiment to existing technology, a short comparison with the Hirst measurements is presented and an issue of the false-positive pollen detections by Rapid-E is discussed.

Detection and Microscopy of <em>Alnus glutinosa</em> Pollen Fluorescence Peculiarities

“…For more than fifteen years there have been attempts to develop semi-automatic [1,15] or automatic [7,10,16] pollen identification systems, in which the use of the fluorescence method for identification of air origin particles occupies an important place. However, to date, the results of collected data, gained by automatic real-time devices, differ [10][11][17][18], and these differences may be determined not only by peculiarities of devices but also by conditions of the environment affecting identified pollen [19][20][21][22]. This fact is confirmed by scientific studies that are targeted at the long-distance pollen transport analysis [23][24][25][26][27][28][29].…”

“…Weather conditions (data taken from Lithuanian hydrometeorological service Šiauliai weather station) in 2018 and in 2019 from January until collection of catkins were different ( Table 1). The beginning of 2019 was colder and the average air temperature in February and March was higher than in 2018. to Šaulienė et al [11]. Until the start of the experiment in 2018, clean, impurity-free pollen was stored in the refrigerator at temperature of 5±1°C and relative humidity of 57±10%; and in 2019, was kept indoors at 25±5°C and 45±10%, respectively.…”

“…Fluorescence of particles is excited using the UV laser (320 nm light) [3]. Ozone-exposed black alder pollen and those not affected by ozone was blown into the device using the method described by Šaulienė et al [11]. The experiment was carried out with Alnus glutinosa pollen whose average size was less than 30 µm [43].…”

“…Studies on pollen in geological samples confirm that the exine of a pollen grain is naturally autofluorescent and the strength of the fluorescence varies with exine thickness [6]. Fluorescence microscopy turned into a tool for studying different pollen type morphology, while acquired knowledge became valuable in analysing not only fossil biological origin particles but also airborne particles [7][8][9][10][11].…”

Tūrisma galamērķu pārvaldības problēmas Latvijā

“…Therefore, the primary attention currently is put to forecasting models that do not use observations in daily routine, being only calibrated and evaluated against them in an offline mode. Automatic pollen monitoring trials from different producers have begun at several European monitoring stations (Scheifinger et al 2013), but for the time being, its accuracy is far behind the manual monitoring accuracy (Crouzy et al 2016;Šauliene et al 2019). Scientists are continuously looking for automatization of the aerobiological monitoring and data collection.…”