Hyper-dimensional computing for enhanced label-free particle analysis in a flow-based optical detection system.

Area of Science:

• Analytical Chemistry
• Biophysics
• Computational Science

Background:

• Flow-based optical detection is crucial for high-throughput particle analysis in microfluidics.
• Conventional methods often require fluorescent labels or complex imaging, posing limitations like cost and potential cell damage.
• Label-free imaging and brain-inspired computing offer promising alternatives for efficient particle characterization.

Purpose of the Study:

• To develop and validate a label-free particle analysis framework integrating Hyper-Dimensional Computing (HDC) with event-based imaging.
• To demonstrate fast and accurate classification of microparticles using this integrated system.
• To explore the impact of optical enhancements, like diffusers, on classification performance.

Main Methods:

• An event-based camera captured optical interference patterns of microparticles in a microfluidic channel.
• Hyper-Dimensional Computing (HDC) was utilized for efficient post-processing and classification of the captured patterns.
• Ground-glass diffusers were incorporated to enhance optical diversity and improve classification accuracy.

Main Results:

• The integrated HDC and event-based imaging system achieved high classification accuracy for microparticles of different sizes.
• Classification accuracy reached up to 98.67% with an optimized ground-glass diffuser configuration.
• The framework demonstrated efficient classification with low computational overhead.

Conclusions:

• The combination of HDC and event-driven photonic detection provides a feasible approach for compact, label-free microparticle classification.
• This framework offers a promising foundation for analyzing more complex biological and industrial particulate systems.
• The study highlights the potential of label-free, computationally driven optical detection in microfluidic applications.