Machine Learning Based Acoustic Manipulation of Particles and Droplets inside Microfluidic Devices - Academic Dissertation

Aug 5, 2026

This thesis introduces a novel approach to ultrasonically manipulate particles and droplets inside microfluidic devices using machine-learning-based control algorithms and bulk acoustic waves. Existing techniques typically rely on piezoelectric transducers driven at resonance or a limited set of harmonics, requiring careful chip design and accurate acoustic field modelling from calibration or simulations. Such dependencies become difficult with complex channel geometries or unexpected disturbances, and particle-tracking-based field modelling is hindered by clustering and focus issues. The study applies ε-greedy, UCB, and an AMA controller variant for 2D manipulation in two microfluidic chips, demonstrating online adaptation to sudden system changes.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages129

Reading Time5 h 25 min

AccessFree to download

Abstract

Preface

Research period and work setting

Supervision and acknowledgements

Funding and community support

Thesis Overview

Motivation and limitations of existing techniques

Proposed machine-learning control approach

Experimental setup and microfluidic chips

Algorithms studied for online adaptation

KYRIACOS CHRYSANTHOS YIANNACOU Machine Learning Based Acoustic Manipulation of Particles and Droplets inside Microfluidic Devices Tampere University Dissertations 1168 Tampere University Dissertations 1168 KYRIACOS CHRYSANTHOS YIANNACOU Machine Learning Based Acoustic Manipulation of Particles and Droplets inside Microfluidic Devices ACADEMIC DISSERTATION To be presented, with the permission of the Faculty of Medicine and Health Technology of Tampere University, for public discussion in the auditorium FA032 Pieni Sali 1 of the Festia building, Korkeakoulunkatu 8 , Tampere, on 7 February 2025, at 12 o’clock. ACADEMIC DISSERTATION Tampere University, Faculty of Medicine and Health Technology Finland Responsible supervisor and Custos Pre-examiners Associate Professor Veikko Sariola Tampere University Finland Associate Professor J. Mark Meacham Washington University in St. Louis USA Assistant Professor Heikki Nieminen Aalto University Finland Opponent Associate Professor Per Augustsson Lund University Sweden The originality of this thesis has been checked using the Turnitin Originality Check service. Copyright ©2025 author Cover design: Roihu Inc. ISBN 978-952-03-3768-1 (print) ISBN 978-952-03-3769-8 (pdf) ISSN 2489-9860 (print) ISSN 2490-0028 (pdf) [http://urn.fi/URN:ISBN:978-952-03-3769-8](http://urn.fi/URN:ISBN:978-952-03-3769-8) Carbon dioxide emissions from printing Tampere University dissertations have been compensated. PunaMusta Oy – Yliopistopaino Joensuu 2025 PREFACE This work was performed during the period November 2019 to May 2024 in the Biomaterials and Robotics Group of the Faculty of Medicine and Health Technology at Tampere University. Working towards becoming a Doctor of Science (Technology), I have many times faced a lot of challenges and obstacles that at the time complicated the journey and made it challenging and not as straightforward as one would think, but at the end, those obstacles taught me valuable lessons and made the degree a fun experience. In this content, I would like to personally thank at very first my supervisor Assoc. Prof. Veikko Sariola. He has been my supervisor since my MSc thesis, and working with him and learning from him has been a true honor. It was a pleasure and my privilege to be able to complete my thesis under his supervision, with that I want to express my thankfulness and gratitude for his constant support, guidance, and supervision throughout this journey. Further, I would also like to thank the members of the follow up steering group, Prof. Pasi Kallio, and Assoc. Prof. Antti Vehkaoja for their valuable suggestions, motivation, and support. Furthermore, I extend my gratitude to the funding bodies that have provided financial support for my thesis. Specifically, my appreciation goes to the Academy of Finland, Walter Ahlström Foundation, TES Foundation, Kone Foundation, and the Doctoral Programme in Medicine, Biosciences and Biomedical Engineering for their invaluable support during my doctoral studies. Moreover, I wish to express my sincere gratitude to my colleagues and the members of our coffee room community. Dr. Vipul Sharma has been a great mentor throughout his tenure as a postdoctoral researcher, offering me unique perspectivesand methodologies in our projects and ensuring that I benefitted from engaging in a wide array of fields. I also extend my thanks to my colleagues Anastasia Koivikko, Mika Pihjamäki, Vilma Lampinen, and everyone in the coffee room for their camaraderie and support. Lastly, I am eternally grateful to my parents Maria and Anthony Yiannacou for their unwavering support and encouragement, and to my dear wife Anastasiia Yiannacou, who has been a relentless source of support throughout my doctoral journey. Tampere, MAY 2024 ABSTRACT This thesis introduces a novel approach to ultrasonically manipulate particles and droplets inside microfluidic devices using machine learning based

FAQs

What problem does the thesis address in microfluidic acoustic manipulation?

It addresses the limitations of existing methods that depend on precise chip design and accurate acoustic field modelling, which becomes difficult with complex geometries, disturbances, and practical measurement challenges during experiments.

What control algorithms are investigated for 2D manipulation?

The thesis studies ε-greedy, UCB, and a new variant called the acoustic model-based adaptive controller (AMA controller), designed to learn manipulation in an online manner.

How does the proposed approach handle sudden changes in the system?

The algorithms adapt to sudden changes during operation, such as when an air bubble enters the manipulation chamber, while still enabling effective particle and droplet manipulation.

Aug 5, 2026

Upload by Seraphina