Skip to main navigation Skip to search Skip to main content

Deep learning classification of particle depth for defocusing 3d-3c micro-PTV

  • Evan Lammertse
  • , Martin Sauzade
  • , Hongxiao Li
  • , Jun Kong
  • , Eric Brouzes
  • Stony Brook University
  • Georgia State University

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We present a high-resolution micro-particle tracking velocimetry (uPTV) technique based on the defocused imaging of tracer particles via brightfield microscopy to experimentally map microfluidic flows. Our technique provides high-resolution 3D data with a regular optical set-up by shifting the technical burden to image processing. We use TrackMate, a free ImageJ plugin, to perform sub-pixel XY tracking, and we obtain Z data via classification of defocused particle images against a reference library of serially stepwise defocused images. Here, we compare a deep learning model with a traditional cross-correlation approach to perform classification.

Original languageEnglish
Title of host publicationMicroTAS 2020 - 24th International Conference on Miniaturized Systems for Chemistry and Life Sciences
PublisherChemical and Biological Microsystems Society
Pages1296-1297
Number of pages2
ISBN (Electronic)9781733419017
StatePublished - 2020
Event24th International Conference on Miniaturized Systems for Chemistry and Life Sciences, MicroTAS 2020 - Virtual, Online
Duration: Oct 4 2020Oct 9 2020

Publication series

NameMicroTAS 2020 - 24th International Conference on Miniaturized Systems for Chemistry and Life Sciences

Conference

Conference24th International Conference on Miniaturized Systems for Chemistry and Life Sciences, MicroTAS 2020
CityVirtual, Online
Period10/4/2010/9/20

Keywords

  • Deep learning
  • Machine learning
  • Microfluidics
  • Particle tracking
  • PIV
  • PTV

Fingerprint

Dive into the research topics of 'Deep learning classification of particle depth for defocusing 3d-3c micro-PTV'. Together they form a unique fingerprint.

Cite this