Towards real-time identification of trapped particles with UMAP-based classifiers

• Type: Article in International Conference Proceedings Book
• Date: 2022

Authors

Teixeira, J (Author) Other; Rocha, V (Author) Other; Oliveira, J (Author) Other; Jorge, PAS (Author) FCUP; Silva, NA (Author) Other

Conference proceedings International

Title: Journal of Physics: Conference Series

5th International Conference on Applications of Optics and Photonics, AOP 2022

17 July 2022 through 22 July 2022

Indexing

Scopus - 0 Citations

Other information

Authenticus ID: P-00X-Q8H

DOI: 10.1088/1742-6596/2407/1/012043

Abstract (EN): Optical trapping provides a way to isolate, manipulate, and probe a wide range of microscopic particles. Moreover, as particle dynamics are strongly affected by their shape and composition, optical tweezers can also be used to identify and classify particles, paving the way for multiple applications such as intelligent microfluidic devices for personalized medicine purposes, or integrated sensing for bioengineering. In this work, we explore the possibility of using properties of the forward scattered radiation of the optical trapping beam to analyze properties of the trapped specimen and deploy an autonomous classification algorithm. For this purpose, we process the signal in the Fourier domain and apply a dimensionality reduction technique using UMAP algorithms, before using the reduced number of features to feed standard machine learning algorithms such as K-nearest neighbors or random forests. Using a stratified 5-fold cross-validation procedure, our results show that the implemented classification strategy allows the identification of particle material with accuracies up to 80%, demonstrating the potential of using signal processing techniques to probe properties of optical trapped particles based on the forward scattered light. Furthermore, preliminary results of an autonomous implementation in a standard experimental optical tweezers setup show similar differentiation capabilities for real-time applications, thus opening some opportunities towards technological applications such as intelligent microfluidic devices and solutions for biochemical and biophysical sensing. © Published under licence by IOP Publishing Ltd.

Language: English

Type (Professor's evaluation): Scientific