Automated Particle and Cell Phenotyping Using Object Recognition and Tracking Based on Machine Learning Algorithms

Gergely Hantos, Gergely Simon, Matěj Hejda, Anne Bernassau, Marc Phillipe Yves Desmulliez

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Downloads (Pure)

Abstract

Knowledge of the acoustic contrast factor of biological entities can provide useful information regarding defects or cell stages in biological applications of microfluidics. It is also a valuable input in the design of acoustic particle manipulators or sorters. To calculate the contrast factor, the required physical properties can be obtained using contact measurements, but these are not desirable as they can damage particles or cells. In indirect approaches, reference particles are employed and the behavior of the unknown particles or cells is compared with that of the reference particles. Here we propose an image recognition-based framework to automate the entire characterization workflow and obtain acoustic contrast factor without intervention. We use 10 micron diameter polystyrene particles as reference and obtain contrast of 6 and 15 micron particle as a proof of concept. Excellent agreement with expected value within 5% is seen for the 15 micron diameter particles.

Original languageEnglish
Title of host publication2021 IEEE International Ultrasonics Symposium (IUS)
ISBN (Electronic)978-1-6654-0355-9
DOIs
Publication statusPublished - 13 Nov 2021
Event2021 IEEE International Ultrasonics Symposium - Xi'an, China
Duration: 11 Sep 202116 Sep 2021

Conference

Conference2021 IEEE International Ultrasonics Symposium
Abbreviated titleIUS 2021
Country/TerritoryChina
Period11/09/2116/09/21

Keywords

  • acoustic contrast factor
  • machine learning
  • mechanical characterization
  • microjluidics
  • object recognition

ASJC Scopus subject areas

  • Acoustics and Ultrasonics

Fingerprint

Dive into the research topics of 'Automated Particle and Cell Phenotyping Using Object Recognition and Tracking Based on Machine Learning Algorithms'. Together they form a unique fingerprint.

Cite this