Automatic Detection and Recognition of Engineered Nanoparticles in SEM Images

Abstract

Engineered nanoparticles have gained importance in recent years and will do so in the future, but their potential toxicity remains an open question. To better understand their effects on the human body, it is necessary to determine their concentration in ambient air. We propose a method to automatically detect nanoparticles in SEM images and differentiate engineered particles from other particles common in ambient air. The method reached Gmeans of 0.985, 0.779 and 0.820 for the classification against non-engineered particles of silver, titanium dioxide and zinc oxide respectively. This is comparable to manual classification.