Connected Component Labelling in the determination of morphometric features of microplastic particles in samples of different matrices

The study attempted to apply the Connected Component Labeling (CCL) method to optimize microscopic analysis of fish tissues and organs collected from the Baltic Sea. The dissected tissues were subjected to chemical degradation. Quantitative analysis including color and shape determination was performed by optical microscopy, while qualitative analysis was performed by FT-IR spectrometry. Visual analysis was performed using a Motic Zoom SMZ-161-BLED stereoscopic microscope (Motic, Spain) at magnifications ranging from 0.75x to 4.5x. After digital image acquisition, in which each pixel is assigned a coordinate of its position in the image and a color value in a specific color space, the image was filtered using different color intensity thresholds. As a result, the image was obtained in the form of a binary matrix, in which the background pixels are characterized by the value 0, while the object pixels are characterized by the value 1. In the analysis of the binary matrix, the implementation of the Union-Find algorithm in C++ language was used, which enabled automatic differentiation of separated regions, giving them unique labels, determining the area of microplastics, and selected morphometric characteristics of the speckles (object size, center of gravity, etc.). A comparison of the values of morphometric characteristics determined algorithmically with those obtained from direct measurements using microscope software showed that the approximate values of the length and width of the microplastic can be effectively estimated by determining the maximum (r=0.88) and minimum ferret (0.77), respectively, and the longer (r=0.84) and shorter (r=0.83) axis of an ellipse with an area analogous to that of the microplastic. The measure of the area of the speck can be successfully estimated using the area of the smallest rectangle (r=74) or the smallest ellipse (r=0.74). Also see: https://micro2024.sciencesconf.org/556544/document