Automated counting of bacterial colonies by image analysis Pei-Ju Chiang a,b , Min-Jen Tseng c , Zong-Sian He a,b , Chia-Hsun Li a,b a Department of Mechanical Engineering, National Chung Cheng University, Chia-Yi, Taiwan b Advanced Institute of Manufacturing with High-Tech Innovations, National Chung Cheng University, Chia-Yi, Taiwan c Department of Life Science, National Chung Cheng University, Chia-Yi, Taiwan abstract article info Article history: Received 4 August 2014 Received in revised form 14 November 2014 Accepted 14 November 2014 Available online 21 November 2014 Keywords: Automated colony counting Image analysis Petri dish Research on microorganisms often involves culturing as a means to determine the survival and proliferation of bacteria. The number of colonies in a culture is counted to calculate the concentration of bacteria in the original broth; however, manual counting can be time-consuming and imprecise. To save time and prevent inconsis- tencies, this study proposes a fully automated counting system using image processing methods. To accurately estimate the number of viable bacteria in a known volume of suspension, colonies distributing over the whole surface area of a plate, including the central and rim areas of a Petri dish are taken into account. The performance of the proposed system is compared with verified manual counts, as well as with two freely available counting software programs. Comparisons show that the proposed system is an effective method with excellent accuracy with mean value of absolute percentage error of 3.37%. A user-friendly graphical user interface is also developed and freely available for download, providing researchers in biomedicine with a more convenient instrument for the enumeration of bacterial colonies. © 2014 Elsevier B.V. All rights reserved. 1. Introduction Bacterial growth is an essential indicator in many studies on micro- organisms. The selection of antibiotics (Van Doorn et al., 2000), toxicol- ogy tests (Chen et al., 2004), and the evaluation of food and drug safety (Itoh et al., 1998) require the determination of microorganism survival rates to verify research achievements. This usually involves counting the number of bacteria in a unit volume of bacterial broth using various methods, including flow cytometry, spectrophotometry, membrane filtering, and the agar plate method. Flow cytometry (Macey, 2007) combines the use of bacterial properties with various fluorescent substances. Bacteria are placed in a flow cytometer, in which the fluorescent substances they carry are excited by lasers set to particular frequencies for the generation of optical signals. Filters of various wave- lengths convert these signals into electronic signals to enable the counting of bacteria. Spectrophotometry (Schmidt and Schmidt, 2004) is a quantitative measurement of the optical transmission of a bacterial suspension as a function of wavelength. The amount of light that passes through the suspension is indicative of the concentration of certain bacteria that do not allow light to pass through. The membrane filter method (Inatomi, 2003) involves passing suitably diluted samples through a membrane filter with pore diameters smaller than those of the microorganisms. The microorganisms remain on the membrane, which is then placed on a culture medium. The total number of bacteria in the original sample can then be calculated according to the number of colonies that form on the membrane filter. The agar plate method (Barbosa, 1995) involves smearing the diluted bacterial suspension on an appropriate culture medium. Since only surviving microbes grow and form colonies on the plate, by counting the number of colonies, the number of viable bacteria can be obtained. Of these methods, the agar plate method is commonly used to assay the survival rate of microbes. However, the manual counting of colonies is time-consuming and imprecise. To save time and prevent inconsistencies, a number of image processing software programs, such as ImageJ, have been devel- oped. ImageJ is a freeware image analysis program that can be used for many image processing and analysis. However, for users who do not have deep knowledge of the image processing, more efforts and familiarity with the language are required to obtain satisfactory results. In addition, Clarke et al. (2010) proposed a low-cost, high-throughput colony counting system consisting of colony counting software and a consumer-grade digital camera or document scanner. The software, called “NICE” (NIST's Integrated Colony Enumerator), reads standard image formats, and therefore may be used in conjunction with many imaging systems. The program (OpenCFU) created by Geissmann (2013) that provides control over the processing parameters can also be used to count cell colonies and other circular objects. Niyazi et al. (2007) developed Clono-Counter, which uses three parameters, namely gray levels, maximum size of one colony, and gray level distribution within the colony, for colony counting. Users need to have some experi- ence to find suitable parameters, but some guidelines are provided to speed up the process. Zhang and Chen (2007) proposed an automatic colony counter for bacterial colony enumeration without any human in- tervention, which has been proven to be more accurate than Clono- Counter. Although it has high accuracy in images with colored media, it has problems with those with transparent media. Chen and Zhang Journal of Microbiological Methods 108 (2015) 74–82 http://dx.doi.org/10.1016/j.mimet.2014.11.009 0167-7012/© 2014 Elsevier B.V. All rights reserved. Contents lists available at ScienceDirect Journal of Microbiological Methods journal homepage: www.elsevier.com/locate/jmicmeth