Biofilm viability checker: An open-source tool for automated biofilm viability analysis from confocal microscopy images

Abstract

Quantifying biofilm formation on surfaces is challenging because traditional microbiological methods, such as total colony-forming units (CFUs), often rely on manual counting. These are laborious, resource intensive techniques, more susceptible to human error. Confocal laser scanning microscopy (CLSM) is a high-resolution technique that allows 3D visualisation of biofilm architecture. In combination with a live/dead stain, it can be used to quantify biofilm viability on both transparent and opaque surfaces. However, there is little consensus on the appropriate methodology to apply in confocal micrograph processing. In this study, we report the development of an image analysis approach to repeatably quantify biofilm viability and surface coverage. We also demonstrate its use for a range of bacterial species and translational applications. This protocol has been created with ease of use and accessibility in mind, to enable researchers who do not specialise in computational techniques to be confident in applying these methods to analyse biofilm micrographs. Furthermore, the simplicity of the method enables the user to adapt it for their bespoke needs. Validation experiments demonstrate the automated analysis is robust and accurate across a range of bacterial species and an improvement on traditional microbiological analysis. Furthermore, application to translational case studies show the automated method is a reliable measurement of biomass and cell viability. This approach will ensure image analysis is an accessible option for those in the microbiology and biomaterials field, improve current detection approaches and ultimately support the development of novel strategies for preventing biofilm formation by ensuring comparability across studies.

Fig. 1. Image analysis steps used in ImageJ to calculate bacterial viability from a confocal image of biofilm with LIVE/DEAD stain.

Images taken from a representative S. sanguinis biofilm cultured for 48 h (20 µm scale bar). See Supplementary Information to implement the automated analysis.

Fig. 2. Results of validation of automatic image analysis protocol.

a ROC curve demonstrating sensitivity and specificity of the image analysis protocol. Green points represent sensitivity and specificity of the green channel (total cells) and red points represent sensitivity and specificity of the red channel (dead cells). b Comparison of image analysis and biological methods. Figure shows mean ± standard deviation (for image analysis, five confocal images were analysed of each of five biological replicates, N = 5 and for biological methods, three biological replicates were analysed, N = 3). To obtain the percentage viability using biological methods, live cells were counted using a serial dilution and CFU-plating. Total cell count was obtained using a haemocytometer. c–f Sample images of a variety of single-species biofilms demonstrating result of automated image analysis. The green outline indicates the total bacteria area and the magenta outline indicates the dead bacteria area. c S. sanguinis (10 µm scale bar), d P. aeruginosa (5 µm scale bar), e multi-species biofilm consisting of F. nucleatum, A. naeslundii, S. gordonii and P. gingivalis (10 µm scale bar), f L. casei (10 µm scale bar). g Representative micrograph of an S. sanguinis biofilm treated with 5% CPC to demonstrate the ability of the macro to handle extreme conditions (Full image 20 µm scale bar, small image 10 µm scale bar). The magenta line shows the result of the segmentation of the red channel. The resulting output from the macro is 0% viability.

Fig. 3. Translation of image analysis method to research applications.

a, b Simple mouthwash study comparing biofilms of a P. aeruginosa and B S. sanguinis treated with mouthwash or water (n = 3 for all conditions). c, d Analysis of S. epidermidis biofilms grown on additively manufactured coupons at different sloping angles: c Percentage alive and d Percentage coverage. e, f Biofilm coverage and viability with increasing distance from coverslip for e a 24-h biofilm of S. sanguinis and f 7-day biofilm of S. sanguinis. Z-stacks were taken at 1 µm increments from the surface (the first plane in which bacteria were identified), and hence the distance from the surface is equivalent to the biofilm thickness.