In February 2021, NSU scientists published a review of the characterizations of single particles using light scattering methods in “Laser and Photonics Reviews”. The review covers research on a wide variety of microparticles, from atmospheric dust to blood cells. The article’s objective is to present developments from various groups around the world. The authors, Andrey Romanov and Maxim Yurkin, work at the Novosibirsk State University Physics Department Laboratory of Structure and Functional Properties of Molecular Systems and the V.V. Voevodsky Institute of Chemical Kinetics and Combustion (ICKC). The review is devoted tomethods for measuring and characterizing single particles, focusing on the latter aspect.
Andrey Romanov, NSU Physics Department postgraduate student and researcher at ICKC SB RAS talked about the review,
While working in this field, we noticed that other scientists do not pay enough attention to research methods. They primarily focus on the object, so they often to some degree reinvent the wheel, repeating a path already traveled by others. Our Cytometry and Biokinetics Laboratory at the ICKC SB RAS, faced a similar problem, so we decided to explore this issue. After an initial acquaintance with the literature on this topic, it was obvious that there are many articles (several hundred) and yet no general understanding.
That is how the idea to write a detailed review for the world community was born. There are many existing methods of characterization, but simply listing them is of little value. The biggest challenge was to compare the methods with each other since each of them is based on a specific experimental setup. This determines the quantity and quality of the received light scattering information . By systematizing and analyzing instrumental approaches to the isolation of single particles and their measurement, we were able to divide the characterization methods into three large classes: data-driven methods, reconstruction, and model methods. The first of these classes is primarily based on machine learning methods to classify on the basis of previously measured data (i.e. how close the new signals are to a previously measured signal). The second class is similar to tomography, i.e., it potentially restores the entire shape of the object. This is, of course, the best option, the “holy grail” in the field of characterization, but the possibilities for existing approaches are severely limited for smaller particles (compared to wavelength). The review pays the most attention to the last class. On the one hand, these methods are limited to the use of a previously known parametric model, on the other these methods are wholly quantitative. They determine not only the characteristics of the object, but the error in these measurements. We previously used these model methods to characterize blood cells.The review’s primary result is the classification description. We believe this is the first outline for research in this field. In particular, we were able to separate it from the much larger , but still qualitative sphere of microscopy (imaging methods in general).
The review presented in this article will help the world community use each other's ideas and developments in the field of single particle measurement and characterization, as well as to develop new approaches for characterization methodology. In the near future, the authors will implement one of these promising paths that is associated with the use of neural networks. The relevance of the review is demonstrated by the high impact factor (10) of the Journal that published it.