Elsevier

Micron

Volume 42, Issue 8, December 2011, Pages 911-920
Micron

Semi-automated Acanthamoeba polyphaga detection and computation of Salmonella typhimurium concentration in spatio-temporal images

https://doi.org/10.1016/j.micron.2011.06.010Get rights and content

Abstract

Interaction between bacteria and protozoa is an increasing area of interest, however there are a few systems that allow extensive observation of the interactions. A semi-automated approach is proposed to analyse a large amount of experimental data and avoid a time demanding manual object classification. We examined a surface system consisting of non nutrient agar with a uniform bacterial lawn that extended over the agar surface, and a spatially localised central population of amoebae. Location and identification of protozoa and quantification of bacteria population are performed by the employment of image analysis techniques in a series of spatial images. The quantitative tools are based on intensity thresholding, or on probabilistic models. To accelerate organism identification, correct classification errors and attain quantitative details of all objects a custom written Graphical User Interfaces has also been developed.

Highlights

► We perform image analysis techniques to segment a large dataset of images with amoebae/cyst population in bacteria lawn. ► Circularity index and probabilistic models are employed to distinguish amoebae/cysts and mobile/immobile organisms. ► A graphical user interface is developed that comprises clustering algorithms and viewing facilities to enhance classification efficiency. ► Probabilistic methods are presented to allow enumeration of bacteria.

Introduction

The importance of interactions between bacteria and protozoa has gained significance since the discovery that Legionella pneumophila could replicate within Acanthamoeba polyphaga (Edelstein and Meyer, 1984, Stevens and O’Dell, 1973). L. pneumophila is an organism which is strongly linked to the development of the Legionnaires’ disease in humans (Edelstein and Meyer, 1984). The recognition of a connection between A. polyphaga and L. pneumophila (Stevens and O’Dell, 1973) has attracted much attention and possible association to other infections has also been investigated (Huws et al., 2008).

A powerful approach to the elucidation of the protozoan–bacteria interactions is to analyse the movement of protozoa in a bacterial lawn. Principally, the movement and behaviour is pertinent to the environment through two main processes, encystment under poor resources and bacteria dependent motility. A statistical analysis of the movement allows determination of the factors that characterise the underlying processes but usually a large number of organisms are necessary to infer useful information and draw a convincing picture. Video sequences that contain such organisms should be analysed to probe the protozoa behaviour (Gaze et al., 2003), however, the large number of the organisms complicates the process towards a quantitative investigation. A variety of sophisticated methods based on neural networks (Casasent and Smokelin, 1994), wavelet transforms (Casasent et al., 1992) and genetic algorithms (Kim and Jung, 2004), have also been widely employed and those state of the art techniques are designed to detect and analyse more complex images. The implementation of algorithms which are based on the above techniques are usually computationally more demanding. Moreover, subdivision of the original images into its constituent parts requires the application of filters (Gerlich et al., 2003), the employment of confinement trees and connected operators (Soille, 1999) or edge-based segmentation (Tvarusko et al., 1998, Tvarusko et al., 1999). By contrast, image analysis techniques based on image thresholding offer an adequate object detection method in pictures or video sequences characterised by a small number of objects with well defined boundaries. Several computer-assisted systems have been developed to track objects of biological importance in video sequences based on intensity thresholding (Degerman et al., 2009, Rabut and Ellenberg, 2004, Sbalzarini and Koumoutsakos, 2005, Tsibidis and Tavernarakis, 2007). With respect to the detection of protozoans in lawns of bacteria, computational tools are required to be capable to facilitate object extraction, provide efficient classification (i.e., cysts or amoebae), determine quantitative details about the organism motility, count the bacterial concentration and offer a correlation of the bacterial distribution to the presence of organisms in the neighbourhood. Although most quantitative tools manage to extract organisms adequately and they are capable to identify protozoa in a series of images, they do not provide a satisfactory answer to how bacteria concentrations could be counted. Moreover, existing algorithms fail to characterise organism motility in two, temporally separated, images by a small time difference if the object boundaries are nearly immobile. Furthermore, specialised image processing methods are not generally applicable to the bacterial counting due to the difficulty to detect bacteria. Most commonly, the intensity of pixels in an image occupied by bacteria is very close to that associated with background which complicates further the bacteria identification. As a result, bacteria enumeration would be a rather laborious process that requires a highly skilled expert. Previous works considered fluorescently tagged bacteria, a method that simplified the identification (Drozdov et al., 2006, Pernthaler et al., 2003, Singleton et al., 2001, Soll et al., 1988). Quantitative tools that address the above issues entail the modification of the existing algorithms with additional routines. The improved algorithms need firstly to manage the detection of the organisms and secondly allow an accurate classification of the two types of organisms in an efficient and robust way.

In the present paper, to overcome existing problems, a semi-automated system has been developed aimed to assist the extraction of quantitative information of A. polyphaga and cysts in a lawn of Salmonella typhimurium bacteria. Image processing techniques are employed to identify objects in a series of images based on thresholding. An automated classification of amoeba and cysts is firstly performed by considering a circularity criterion (i.e., cysts are almost circular) and image segmentation is based on intensity thresholding. A semi-automated correction of the object distinction is subsequently performed by means of a Graphical User Interface (GUI) designed in Matlab (The Math Works, Natick, MA) which provides a synergy of clustering and viewing tools to accelerate organism identification. The method aims to integrate image processing and statistical classification technologies. A synergy of image processing techniques and probabilistic models for noise are subsequently used to assist in the bacterial counting. The proposed method is used as an alternative to the employment of fluorescent bacteria as a technique to facilitate bacterial detection. A similar approach is ensued towards the characterisation the organism motility. The algorithms developed in this paper intend to be an essential component of a completely automated system that will allow an adequate quantification of protozoa–bacteria interactions.

Section snippets

Strains

The virulent strain of S. typhimurium was used in co-culture experiments, and was grown in Luria Broth (LB). A. polyphaga CCAP 1501/18 was used as a model organism since L. pneumophila has been shown to grow in this strain. A. polyphaga was grown axenically in proteose peptone glucose medium (PPG).

Co-culture experiments

Bacterial cultures were washed 3 times in Page's Amoeba Saline (PAS) and diluted to approximately 108 cells/ml (high bacteria concentration) and 107 cells/ml (medium bacteria concentration). 100 ml were

Results and discussion

Basic image analysis techniques based on intensity thresholding were firstly used to segment all set of images and subsequently locate and count amoebae and cysts. Image processing algorithms were applied to all datasets aimed to identify all objects and determine a set of morphological details such as position, shape, perimeter and area. Small objects of no importance were removed from consideration by adapting the algorithm to ignore tiny, light regions (i.e., comprising fewer than 20

Conclusions

We have developed a widely applicable technique for accurate analysis and elucidation of amoebae–bacteria interactions. In the first application, amoebae and cysts were localised on lawns of a varied bacterial density and morphological quantities were obtained. The method proved to be particularly useful because it allowed a semi-automated analysis with a remarkably high efficiency and reliability. A second technique was additionally presented that aimed to compute the bacterial coverage in

References (24)

  • F. Ekelund et al.

    Population dynamics of active and total ciliate populations in arable soil amended with wheat

    Appl. Environ. Microbiol.

    (2002)
  • W.H. Gaze et al.

    Interactions between Salmonella typhimurium and Acanthamoeba polyphaga, and observation of a new mode of intracellular growth within contractile vacuoles

    Microb. Ecol.

    (2003)
  • Cited by (0)

    View full text