Back to search

National Technical University of Athens (NTUA)

Τεχνικές με femtosecond lasers και νευρομορφικές κάμερες συμβάντων για απεικόνιση στην κυτταρομετρία ροής και την ταξινόμηση μικροσωματιδίων

Abstract

dc:description

Imaging flow cytometry combines the statistical throughput of flow cytometry with the spatial information of microscopy. Its extension to high-flow regimes is constrained by two coupled bottlenecks: the generation of large volumes of redundant image data and the motion blur that arises when particles move appreciably during sensor exposure. This doctoral dissertation experimentally investigates photonic and neuromorphic imaging architectures that address these limitations through physical encoding, sparse sensing and optical preprocessing. The work focuses on the design, construction, alignment, characterization and experimental operation of high-speed imaging flow cytometry platforms, together with the acquisition, organization and physical interpretation of the resulting datasets. First, a time-stretched imaging flow cytometry platform was developed and characterized. In this architecture, spatial information from flowing microparticles is mapped onto a temporal waveform through spectral and dispersive encoding. This approach enables high-rate optical readout using fast photodetection rather than conventional frame-based image acquisition, while exposing the principal experimental constraints associated with optical loss, signal-to-noise ratio and coupling stability. In a time-stretched imaging cytometry experiment, temporal traces from PMMA microparticles were used for two-class particle classification at an imaging throughput of up to 62,000 particles per second. Second, an event-based imaging flow cytometry platform was implemented experimentally. In contrast to conventional cameras, event-based sensors do not acquire full frames at fixed intervals. Instead, each pixel independently emits an asynchronous event when the local logarithmic change in brightness exceeds a predefined contrast threshold. This sensing principle transfers part of the data reduction process to the detector itself, suppresses the redundancy of the static background and enables high-temporal-resolution recording of flowing PMMA microparticles with reduced motion blur. The datasets generated by these platforms were used to evaluate digital and photonic classification architectures. Photonic preprocessing was based on Optical Spectrum Slicing, in which optical filters partition and weight spectral components of the signal before the digital backend. In the event-based pathway, coupling the neuromorphic camera to photonic preprocessing enabled PMMA microparticle classification with an accuracy of 98.6%, while reducing the number of trainable parameters in the digital network by up to a factor of 22. In the time-stretched pathway, Optical Spectrum Slicing improved classification accuracy from 87.8% to 91% and reduced the complexity of the digital backend by a factor of 2.5. Together, these results show that high-speed imaging flow cytometry can be formulated as a co-design problem involving optical acquisition, physical encoding, sparse sensing and efficient classification. Time-stretched imaging provides ultrafast photonic readout; event-based sensing reduces data redundancy at the detector level; and photonic preprocessing decreases digital complexity while preserving task-relevant information. The dissertation therefore establishes an experimental basis for next-generation imaging cytometry systems in which measurement and preprocessing are treated as components of a unified sensing–processing pipeline, with potential extension to real-time operation, biological samples and active closed-loop sorting.

Degree

thesis:*
Grantor dc:publisher
National Technical University of Athens (NTUA)
Year dc:date
2026

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Tsilikas, Ioannis
  • Τσιλίκας, Ιωάννης

Subjects

dc:subject × 17

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/62162

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Tsilikas, Ioannis; Τσιλίκας, Ιωάννης. Τεχνικές με femtosecond lasers και νευρομορφικές κάμερες συμβάντων για απεικόνιση στην κυτταρομετρία ροής και την ταξινόμηση μικροσωματιδίων. National Technical University of Athens (NTUA), 2026. http://hdl.handle.net/10442/hedi/62162