Estimating the concentration of gold nanoparticles incorporated on natural rubber membranes using multi-level starlet optimal segmentation

Nenhuma Miniatura disponível
Siqueira, A. F. de [UNESP]
Cabrera, F. C. [UNESP]
Pagamisse, A. [UNESP]
Job, A. E. [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
This study consolidates multi-level starlet segmentation (MLSS) and multi-level starlet optimal segmentation (MLSOS) techniques for photomicrograph segmentation, based on starlet wavelet detail levels to separate areas of interest in an input image. Several segmentation levels can be obtained using MLSS; after that, Matthews correlation coefficient is used to choose an optimal segmentation level, giving rise to MLSOS. In this paper, MLSOS is employed to estimate the concentration of gold nanoparticles with diameter around 47 nm, reduced on natural rubber membranes. These samples were used for the construction of SERS/SERRS substrates and in the study of the influence of natural rubber membranes with incorporated gold nanoparticles on the physiology of Leishmania braziliensis. Precision, recall, and accuracy are used to evaluate the segmentation performance, and MLSOS presents an accuracy greater than 88 % for this application.
Computational vision, Gold nanoparticles, Image processing, Multi-level starlet segmentation, Natural rubber, Scanning electron microscopy, Wavelets, Modeling and simulation
Como citar
Journal Of Nanoparticle Research. Dordrecht: Springer, v. 16, n. 12, 13 p., 2014.