On the robustness of machine learning algorithms toward microfluidic distortions for cell classification <i>via</i> on-chip fluorescence microscopy
Ali Ahmad, Federico Sala, Petra Paiè, Alessia Candeo, Sarah D’Annunzio, Alessio Zippo, Carole Frindel, Roberto Osellame, Francesca Bragheri, Andrea Li Bassi, David Rousseau
Abstract
-projection transformation, to reach an efficient real time classification with an accuracy of 99.4% using a convolutional neural network with transfer learning from simulated data. Beyond this specific use-case, we provide a web platform to generate a synthetic dataset and to investigate the effect of flow speed on cell classification for any biological samples and a large variety of fluorescence microscopes (https://www.creatis.insa-lyon.fr/site7/en/MicroVIP).
Topics & Concepts
MicrofluidicsComputer scienceRobustness (evolution)Artificial intelligenceFluorescence microscopeConvolutional neural networkMicroscopyMotion blurAlgorithmMachine learningPattern recognition (psychology)Computer visionFluorescenceNanotechnologyMaterials scienceChemistryOpticsImage (mathematics)PhysicsBiochemistryGeneMicrofluidic and Bio-sensing TechnologiesCell Image Analysis TechniquesDigital Holography and Microscopy