Zeroing in on digital imaging in microbiology

March 19, 2014

Digital imaging has revolutionized many industries and many applications that touch our lives today. For example, a patient can go to the hospital to get an x-ray; by the time the patient gets to the office of the primary care physician, the doctor already has the x-ray on the screen and is ready to do the analysis and discuss the results. Before digital x-ray technology, traditional photographic film had been used to review x-rays. With digital technology, not only is time saved bypassing the physical development of the traditional film, but there is the benefit of being able to enlarge the digital image for better analysis. This same concept of moving from traditional analysis of results to digital imaging for analysis is starting to occur across today’s Microbiology labs in North America.

Why the move to digital imaging?

What is motivating the sudden upsurge in interest in digital image analysis in Microbiology? Just a few years ago, the biggest breakthrough in the field was front-end automated specimen processing of planting and streaking. Now, the Microbiology community seems to have shifted from a passing interest in front-end automation to a large-scale interest in full laboratory automation and digital microbiology. Microbiology is facing an impending crisis for many well documented reasons, including those given by Novak and Marlow: “budget cuts, shrinking workforce, and legislation-mandated testing.”1 The impact that the Affordable Care Act will have on the workload volume of laboratories across the United States will have is not yet known, but many laboratory professionals are already wondering how they can do more with less and still maintain high levels of patient care. 

The timing of technology is a paramount reason for the shift of interest toward full automation and digital imaging in Microbiology. It is always a priority to improve efficiency and processes, but sometimes one has to wait for the technology to come along in a format that is user-friendly and available. An analogy to the new wave of interest in automation and digital imaging is bacterial identification using mass spectrometry. That technology has been available for decades; an example of its use can be found in an academic publication dating back to 1975, proposed by John P. Anhalt and Catherine Fenselau.2 However, only recently, as a number of companies have made the technology available in a user-friendly format, has its adoption in laboratories across the United States greatly increased. 

Similarly, progressive manufacturers have only recently made the technology for full laboratory automation and digital imaging in Microbiology scalable and user-friendly. Also, now that the technology is available in the U.S. to improve efficiency and reduce turnaround times, the interest has spiked. Multiple articles have been written recently to explain automated specimen processing and full laboratory automation.1,3,4 Fully automated and digital Microbiology systems consist of front-end specimen processors and conveyor track connecting to smart incubators that include image acquisition stations for subsequent analysis. This article zeroes in on the key factor that will impact digital imaging and image analysis as Microbiology laboratories make the transition to full automation. 

Time zero: the key concept

Any researcher will emphasize that time zero is crucial to good results in scientific studies. For example, in epidemiological research, the issue of time is central to, in Kraemer’s words, “affecting sampling, measurement, design, analysis and, perhaps most important, the interpretation of results that might influence clinical and public-health decision making and subsequent clinical research.”5 In Microbiology, the issue of time is no different than in epidemiological research; time zero creates the start point to later compare results against the baseline. When moving from traditional microbiology to full laboratory automation, the use of smart incubators and image acquisition stations enters the picture, and image acquisition at time zero becomes the pillar of successful image analysis. 

In Microbiology, aside from the unavailability until fairly recently of the technology, another obstacle for automation and the migration to digital technology is, in Bourbeau and Ledeboer’s words, “the perception that machines cannot exercise the critical decision-making skills required to process microbiology specimens….Specifically, human observation of organism growth on agar plates is still considered essential by many.”3 In reality, automation does not intend to take humans out of the equation, but to provide and present data in a format that makes it easy to screen and to create reports that are faster to manage. 

Moving from physically looking at a culture plate to looking at a screen is a big change. That is because laboratory personnel reading the plates move them to achieve adequate lighting depending on the type of plate or organism they are looking for (e.g., they, raise the plate against light to see hemolytic cultures). To mimic what humans do, microbiology automatic image acquisition stations need to capture high-quality images of culture plates, using different lighting systems. In addition to capturing the images using the correct lighting combination, automatic image acquisition stations need to do so immediately after the plates are received in the smart incubator to record any existing plate precipitates or specimen deposits existing on the agar prior to incubation. Time zero image is important because it creates a baseline to set the rules for image analysis. These rules include algorithms to detect growth, count colonies, or detect colony colors following culture plate incubation. 

Illustration 1. Charlie Chaplin demonstrates differential image analysis.

Humans are flexible and can make decisions and discern. Instruments are clever, but only because they have rules. Having a time zero image of the culture plate allows the system’s smart software to use its algorithms to compare the existing image to the baseline to start using rules for interpretation. In a sense, comparative differential image analysis in Microbiology has built-in artificial intelligence (AI) derived from time zero. To illustrate the point of comparative differential image analysis, please refer to Illustration 1. In this analogy, Charlie Chaplin is equivalent to a culture plate. The image analysis system has acquired the image of Charlie Chaplin at time zero, then at 16 hours. In the figure at 0 hours, he does not have a mustache; in the figure at 16 hours, he has a mustache. The way comparative differential image analysis works is by detecting and highlighting only the “new” information, as compared to the time zero image, and processing only what is new.

Image analysis aids decision making

Figure 1. Plate at time zero.

Using these new Image Analysis systems for Microbiology, determining growth or no growth in a culture plate or colony in terms of CFUs is possible thanks to the time zero picture. Using that key image as the baseline, the systems, based on algorithms, can differentiate whether a sample is positive or how positive it is. Figure 1, 2, and 3 show how the system applies its artificial intelligence to detect the existing contamination on the plate and at its predetermined reading time. It ignores the “noise” generated by the contamination of the original plate. Figure 1 was captured at time zero. At the bottom of the plate, one can read the type of media, the expiration date, and the lot number, and there is even a small piece of tape at the top.

Figure 2. Plate with growth.

Figure 2 is the same plate after incubation; this plate has some growth on it. Figure 3 shows how differential image analysis works and how the system views the plate behind the scenes. It is able to ignore the “noise” caused by previously existing elements on the plate, and it can isolate only what is relevant.

Figure 3. Differential image analysis as recognized by software.

Artificial intelligence through algorithms of image analysis is not meant to make decisions for laboratory professionals, but to present data in easy-to-report format that can be managed faster. AI can pre-sort culture plates into groups or categories in a way that helps improve the workflow of the lab. For example, it can group culture plates from highest number of colonies to no colonies, so that the clinical laboratory professional can read and order the workups on the positives first. Alternatively, it can group all “no-growth” urines first so that the technologists can report those negative results as soon as cultures have completed their assigned incubation time. When color recognition capabilities are added, the presence or absence of specific pathogens automatically on chromogenic medium can be detected. Digital imaging eliminates hands-on time and allows for the reading to be done remotely because it is not limited to the confines of the laboratory. This technology brings the promise of improved patient care through reduced turnaround times, changing the way of working to get results back to clinicians faster.  

For medical professionals and patients waiting for results, time zero is also important, and it starts from the moment the sample is collected until the results are back in the hands of the healthcare professional. There is a therapeutic window in which the patient can be treated and, Novak and Marlowe note, “it has been estimated that a majority, or approximately two-thirds, of health care decisions are based on laboratory results.”1 The faster laboratories can turn around results, the sooner the patient can receive the correct medication and treatment. Digital imaging in Microbiology brings the laboratory back to the patient’s bedside by giving the clinician an electronic record of the patient’s culture plate with the colony count and Gram stain imaging along with the laboratory interpretation of the results—accessible on a handheld device like an iPad or tablet. (Illustrations 2 and 3). Both the laboratory and the clinician will be more engaged and involved sharing the results on screen. Clinicians will be able to see the pathogen and view the culture plate or Gram smear, enabling them to complete the clinical picture by then correlating the results with what they are seeing at the bedside. With the laboratory results in a handheld device, the clinicians save the time it takes to call in for results. They can even magnify and scrutinize what they are seeing on the screen. Accessing results in real time reduces the turnaround time. Also, digital imaging increases collaboration between the point of care and the laboratory. 

Illustration 2. Bedside iPad.
Illustration 3. Bedside tablet.

Digital imaging has arrived for Microbiology in North America, and it may be commonplace within the next few years. However, it may not be suited for every laboratory. To return to the x-ray analogy: while many medical facilities have implemented digital x-ray systems, some find that their needs are still best met by using the traditional photographic films. As laboratories struggle with personnel shortages and other issues, and are rethinking the ways they operate to accommodate projected increased volumes, they need to evaluate whether automation and digital imaging in
Microbiology work for their needs, and whether this new technology will help them meet the challenges ahead. 

Gabriela Franco is COPAN Diagnostics’ Director of Marketing. Gabriela has more than 10 years of experience in marketing and product management. She holds a BS in Business Administration and Marketing, and a Master of International Management (MIM) degree.


  1. Novak SM, Marlowe EM. Automation in the clinical microbiology laboratory Clin Lab Med. 2013; 33(3):567–588. 
  2. Anhalt JP, Fenselau C. Identification of bacteria using mass spectrometry. Anal Chem. 1975;47(2):219–225.doi: 10.1021/ac60352a007.
  3. Bourbeau PP, Ledeboer NA. Automation in clinical microbiology. J. Clin. Microbiol. 2013; 51(6):1658-1665. 
  4. Greub G Prod’hom G. Automation in clinical bacteriology: what system to choose? Clin Microbiol Infect. 2011;17(5):655–660.
  5. Kraemer, H. Epidemiological methods: about time. Int J Environ Res Public Health. 2010; 7(1):29–45.