Initial Bacterial Adhesion Counting on Nanopyramid-Structured Surfaces Using Artificial-Intelligence Technique

Abstract

Objectives: Bacterial adhesion is the primarily step in biofilm formation. It is an important factor to evaluate the biological properties of various materials. Measurement of the bacterial adhesion has been of great interest to many research groups from multiple subjects. Many methods have been used for bacterial counting, such as microbial colony counts, Gram staining, or nucleic acid quantitation. Direct measurement provides a directly perceived result, more information can be acquired from the results such as bacteria-material surface interactions and the distribution of bacteria. This study provides a new approach using artificial-intelligence (AI) method to measure directly the initial bacterial adhesion from Scanning Electron Microscope (SEM) images. Methods: Nanopyramid structured Polystyrene (PS) film (Sigma-Aldrich, molecular weight ca. 192,000) was prepared by moulding process. Streptococcus mutans (S.m) was used for the bacterial adhesion on nanopyramid structured surface. Biofilm formation was evaluated at time points of 1 hour, 1 day, 3 days and 7 days respectively. SEM images of bacteria adhered surfaces at 2500× magnification were taken at 3 random areas on each film, 3 films were tested in each time point. SEM image pre-processing and bacterial area measurement were performed using Fiji (ImageJ) software and Trainable Weka Segmentation plugin in Fiji, respectively. Confocal laser scanning microscopy (CLSM) images with live plus dead cell counts on the nanopyramid surface were used as control. Results: Same pattern were found about bacterial adhesion between live/dead staining method using CLSM images and the newly proposed AI method using SEM images (Fig.1). Conclusions: AI method on SEM images can be utilized directly for both morphology and quantity analysis of bacterial adhesion, it can be a new tool to measure accurately the initial bacterial adhesion on nano-structured surfaces.