Particle Classification through the Analysis of the Forward Scattered Signal in Optical Tweezers

Abstract

The ability to select, isolate, and manipulate micron-sized particles or small clusters has made optical tweezers one of the emergent tools for modern biotechnology. In conventional setups, the classification of the trapped specimen is usually achieved through the acquired image, the scattered signal, or additional information such as Raman spectroscopy. In this work, we propose a solution that uses the temporal data signal from the scattering process of the trapping laser, acquired with a quadrant photodetector. Our methodology rests on a pre-processing strategy that combines Fourier transform and principal component analysis to reduce the dimension of the data and perform relevant feature extraction. Testing a wide range of standard machine learning algorithms, it is shown that this methodology allows achieving accuracy performances around 90%, validating the concept of using the temporal dynamics of the scattering signal for the classification task. Achieved with 500 millisecond signals and leveraging on methods of low computational footprint, the results presented pave the way for the deployment of alternative and faster classification methodologies in optical trapping technologies.

Keywords: Brownian motion; optical trapping; optical tweezers; particle identification; principal component analysis.

Figure A1

Results of the PCA analysis of Channels X, Y and $$SUM$$. Left-hand side plots depict the first two principal components obtained for the PCA analysis as well as a typical signal obtained for two distinct particles for the 3 distinct channels. Subfigures on the right-hand side display the results of the PCA analysis as a two-dimensional plot with the first two principal components as the extracted features. Additionally represented and highlighted by the arrows are the signals for the particles represented on the left-hand side plots.

Figure A2

Scree plot for the PCA analysis presented in Figure A1, displaying the explained and accumulated variance ratio in function of the principal components used.

Figure 1

(A) Schematic of the optical tweezers system. (B) Schematic representation of the classification procedure, depicting the time scope of signals acquired from the quadrant photodetector (X, Y, $$SUM$$) and the PCA plots of the corresponding Fourier transforms for all the tested particles.

Figure 2

Confusion matrices showing the classification performance of the tested algorithms. The labels correspond to each particle type with the score corresponding to the mean accuracy obtained for the cross-validation procedure, as described in the main text.

Figure 3

Confusion matrices showing the classification performance of the tested algorithms as in Figure 2. In this case, only the $$SUM$$ channel of the QPD used.