Performance of the Swisens Poleno automatic air-flow cytometer in Nordic conditions

Real-time air-flow cytometry, a technique for analysing microparticles suspended in the air, has advanced rapidly with the development of state-of-the-art automatic monitors. This study presents results obtained using measurements from the Swisens Poleno cytometer, which, in our case, were processed using a pollen recognition algorithm primarily trained to identify species typical of Southern Finland.

The algorithm performed remarkably well under laboratory conditions, achieving an overall accuracy of nearly 90%. Most misclassifications involved pollen grains from species within the same pollen family, an understandable outcome given their morphological similarities. However, applying the classification model to environmental measurements collected under real atmospheric conditions revealed additional challenges. The most significant issue was the occurrence of false positive recognitions, where the algorithm mistakenly identified particles as pollen that could not realistically be present in the air at the time of measurement. This discrepancy was identified through parallel measurements conducted with a manually operated Hirst-type pollen trap. To address this, previously neglected fluorescence signal was integrated into the algorithm, which partially mitigated the issue. We present a comparison of pollen concentrations measured by the Poleno cytometer, using the improved recognition algorithm, and the manual trap across multiple seasons and locations in Northern Europe. While the correlation between the two methods was slightly lower than expected based on laboratory results, it reached approximately 0.85 for the main pollen species in bi-hourly measurement intervals. Additionally, we demonstrate the device’s reliability under the harsh weather conditions of Arctic winters.