Authors: Luca Bertoli, Alice Negri, Davide Melegari - IS E. Fermi, Mantua

QuickSpot aims to create a methodology to increase the speed of investigation of airborne fungal microorganisms, in indoor and outdoor environments, thus reducing the application time required for the recognition process. This is made possible through modern aerobiological sampling techniques and through pattern recognition techniques, based on neural network. The use of these techniques, combined with each other, allows the automation of the entire process and the reduction of the timeframe associated with recognition and analysis.

In order to train the neural network for recognition, it must be instructed with a library of images depicting the pathogenic microorganism in question. The first pathogenic microorganism investigated by QuickSpot, on the basis of its hazard and presence in agriculture and health, is the fungus Alternaria alternata.

The methodology consists of the following steps. First, a system consisting of sensors 

detects atmospheric conditions (temperature and humidity).If these conditions promote the growth of

of pathogens, the operator is instructed to sample the air in the surrounding environment. In

second, the product of this sampling is then analyzed in a laboratory with an

optical microscope. Third, the acquired images are analyzed by a system of

recognition based on neural networks and deep learning. 

For this purpose, we built a customized sterile box to generate libraries on which the neural network is trained. Fourth, the neural network informs field workers of the presence of dangerous pathogens, enabling them to organize direct operational plans with pesticides and/or recurring preventive controls. This innovation will make it easier for field workers to decide on

possible prevention activities in order to avoid damage caused by airborne fungal agents.

Acknowledgements received:

• Participation in Regeneron ISEF 2021 (virtual).