Bacterial Density and Biofilm Structure Determined by Optical Coherence Tomography

Abstract

Optical-coherence-tomography (OCT) is a non-destructive tool for biofilm imaging, not requiring staining, and used to measure biofilm thickness and putative comparison of biofilm structure based on signal intensity distributions in OCT-images. Quantitative comparison of biofilm signal intensities in OCT-images, is difficult due to the auto-scaling applied in OCT-instruments to ensure optimal quality of individual images. Here, we developed a method to eliminate the influence of auto-scaling in order to allow quantitative comparison of biofilm densities in different images. Auto- and re-scaled signal intensities could be qualitatively interpreted in line with biofilm characteristics for single and multi-species biofilms of different strains and species (cocci and rod-shaped organisms), demonstrating qualitative validity of auto- and re-scaling analyses. However, specific features of pseudomonas and oral multi-species biofilms were more prominently expressed after re-scaling. Quantitative validation was obtained by relating average auto- and re-scaled signal intensities across biofilm images with volumetric-bacterial-densities in biofilms, independently obtained using enumeration of bacterial numbers per unit biofilm volume. The signal intensities in auto-scaled biofilm images did not significantly relate with volumetric-bacterial-densities, whereas re-scaled intensities in images of biofilms of widely different strains and species increased linearly with independently determined volumetric-bacterial-densities in the biofilms. Herewith, the proposed re-scaling of signal intensity distributions in OCT-images significantly enhances the possibilities of biofilm imaging using OCT.

Figure 1

2D cross-sectional OCT images of the four case biofilms evaluated. The scale bars indicate 100 µm.

Figure 2

Influence of the OCT focus point on the signal intensity distribution across a 2D cross-sectional image of a homogeneous agar layer. OCT images represent the cross-sectional view of an MRS agar layer at different focus points, indicated by the arrows. For each image, the average signal intensity is presented as a function of the height in the agar layer. Scale bars equal 200 µm. a.u. denotes “arbitrary unit”.

Figure 3

Average intensity (auto-scaled) in OCT images of bacterial suspensions in PBS as a function of the optical density, OD_600, of the suspensions. Data involve different concentrations of widely different strains and species (S. epidermidis 252, S. epidermidis ATCC 35984, P. aeruginosa 39324, S. mutans UA159 or S. oralis J22). Different strains are indicated by different symbols, but not further specified due to overlap of data points at similar optical densities. Drawn lines represent the best fit to an assumed linear relation with correlation coefficient R² and significance of the slope P indicated.

Figure 4

Auto- and re-scaling based analyses of 2D cross-sectional OCT images. (A) The back-scattered light from the measured sample is collected by the OCT camera and its intensity outputted as analogue voltage. Panel A showed an example of the output voltage in the Z-direction perpendicular to the substrate. Note the OCT adjust the level of zero voltage to ensure zero average output over an entire image. (B) In the subsequent auto-scaling process, voltage is firstly expressed in decibel units with respect to a reference intensity provided by the instruments after which the decibel scale is digitized in 256 discrete values from to 0 and 255 a.u. (panel B1) to yield the image provided in panel B2. Intensity distribution in auto-scaled images is created by the OCT instrument to ensure optimal quality of an individual image. (C) Next, Otsu thresholding is applied on the OCT image to determine the biofilm surface (green line), while the substratum surface is visually identified based on the abrupt increase in intensity as compared with the biofilm interior. In order to avoid a potential impact of the intensity of the substratum material on the intensity of the biofilm, for calculational purposes the substratum surface was positioned 3 µm above the visually identified surface (red line). (D) In the proposed re-scaling process of OCT images, the average auto-scaled signal intensity above the biofilm surface, as identified by Otsu thresholding, is given a new intensity value of 0 a.u., while the separately measured, average intensity of a biofilm-free substratum is used and adjusted to an intensity value of 255 a.u. (panel D1). A new OCT image is subsequently generated with the re-scaled signal intensity distribution (panel D2).

Figure 5

Signal intensity analysis of OCT images: biofilms of non-EPS producing S. epidermidis 252 and EPS producing S. epidermidis ATCC 35984. (A) Schematic presentation of non-EPS producing S. epidermidis 252 and EPS producing S. epidermidis ATCC 35984 biofilms (see Table 1) and their measured thickness (derived using Otsu thresholding of auto-scaled OCT images) and volumetric bacterial density, calculated from the biofilm thickness and enumeration of the number of bacteria in the biofilm. (B) Intensity as a function of biofilm height (%) for S. epidermidis 252 and S. epidermidis ATCC 35984 biofilms in auto-scaling analysis. (C) Same as (B), but now in re-scaling analysis. Error bars indicate standard deviations over triplicate experiments with separate bacterial cultures.

Figure 6

Signal intensity analysis of OCT images: S. mutans UA159 biofilms grown in medium with 0.5% or 1% sucrose added. (A) Schematic presentation of S. mutans biofilms grown in medium with 0.5% or 1% sucrose added (see Table 1) and measured thickness and volumetric bacterial density. (B) Intensity as a function of biofilm height (%) for S. mutans biofilms grown with 0.5% or 1% sucrose added in auto-scaling analysis. (C) Same as (B), but now in re-scaling analysis. Error bars indicate standard deviations over triplicate experiments with separate bacterial cultures.

Figure 7

Signal intensity analysis of OCT images of P. aeruginosa ATCC 39324 biofilms grown in LB and ASM medium. (A) Schematic presentation of P. aeruginosa biofilms grown in LB and ASM medium (see Table 1) and measured thickness and volumetric bacterial density. Note that due to growth in a CDFF, thicknesses are identical. (B) Intensity as a function of biofilm height (%) for P. aeruginosa biofilms grown in LB or ASM medium in auto-scaling analysis. (C) Same as (B), but now in re-scaling analysis. Error bars indicate standard deviations over triplicate experiments with separate bacterial cultures.

Figure 8

Signal intensity analysis of OCT images of single-species S. oralis J22 and A. naeslundii T14V-J1 biofilms and multi-species biofilms. (A) Schematic presentation of the biofilms for both single-species and the more compact multi-species biofilms (see Table 1) and measured thickness and volumetric bacterial density. (B) Intensity as a function of biofilm height (%) for both single-species and multi-species biofilms, in auto-scaling analysis. (C) Same as (B), but now in re-scaling analysis. Error bars indicate standard deviations over triplicate experiments with separate bacterial cultures.

Figure 9

Average signal intensity in OCT images of biofilms as a function of the volumetric bacterial density for all individual biofilms grown of the different cases in auto- (panel A) and re-scaling analysis (panel B). Drawn lines represent the best fit to an assumed linear relation with correlation coefficient R² and significance of the slope P indicated. Dotted lines represent 95% confidence intervals.