Event Driven Automated Microscopy

Applications in Cancer Research

  • Fig. 1: Workflow for event driven microscopy. Images are analyzed as they are acquired and used to influence subsequent acquisition steps. Ideally a two way interface between the acquisition software and external image analysis packages should also be established. The tools used for our implementation, including the Zeiss Open Application Development (OAD) framework, are shown in blue.
  • Fig. 2: Search and find assay for metaphase spreads. A low resolution scan with a 20X air objective is performed across the entire slide. Low resolution images are analyzed to detect the position of spreads. After switching to a 1.4 NA oil objective, high resolution z-stacks are captured at the detected positions. ­
  • Fig. 3: High resolution live cell imaging of mitosis. a) Using a predefined search set, typically 20 positions, bright-field (or phase-contrast) images are acquired 7.5µm above the focal plane as determined by autofocus. These images are acquired, and analyzed, until a new mitotic cell is detected. In this case, a multi-channel 3D fluorescent time-lapse is performed. As this position will have now been exposed to a significant quantity of light it is deleted from the search set and a new position is added. This process is repeated over 24 hours. It is simple to adapt this protocol for multi-well experiments.  b) Example bright-field image with detected mitotic cells shown by red circles. Detection is facilitated by the circular Hough transform which identifies the cells as they round up. The first 4 time-points of the associated fluorescence time-lapse are shown. These images are maximal projections of data which was deconvolved using Huygens Professional (Scientific Volume Imaging).

Why Do Event Driven Microscopy? In a conventional micros­copy study the researcher defines an experimental configuration, typically a combination of time-lapse, z-stack, multi-channel, multi-position and multi-view settings. Importantly, everything the microscope will do is fully defined before the experiment is initiated. In an event driven approach sample variation is used to influence the experimental configuration. This can be done with closed feedback loops where image analysis protocols are run on images as they are acquired, and the results of this analysis are used to decide what happens next (fig. 1). These approaches have the potential to save a vast amount of user time, enabling larger scale studies, and a reduction in user error and bias. Feedback loops can also be utilized to reduce the chance of experimental failure. This can be as simple as correcting for stage drift or following a moving target if it leaves the field of view. Modern commercial microscopes can acquire a large, perhaps unmanageable, amount of data. In many studies only a fraction of this data contains useful information which can be used to answer a specific biological question. An event driven approach can be used to filter images as they are acquired and dramatically reduce the data storage requirements of a study. For example, if we are studying a mitotic phenotype using time-lapse microscopy then only the regions of the images containing mitotic cells should be saved for further analysis. In light microscopy there is a constant trade-off between spatial resolution, frame rate, signal to noise ratio (SNR) and light exposure. When imaging live samples a limited, ideally physiological, light exposure should be used to ensure the relevant biology is not affected by the imaging. An alternative use of event driven microscopy is to reduce the sample light exposure by only imaging what is needed. This intelligent use of the light budget allows for a combination of higher SNR, spatial resolution and frame rate with equivalent total light exposure.


In a recent trend the acquisition software for many wide-field and confocal systems can perform simple event driven approaches.

To facilitate generalized and complex applications it is our opinion that acquisition software should satisfy two key requirements. Firstly, control of fundamental microscope functions such as stage movement, objective change and experimental configuration should be fully scriptable, and have thorough documentation. To enable the use of cutting edge image analysis the use of flexible external platforms such as Fiji, Icy,  CellProfiler or Matlab is advantageous [1-3]. Therefore our second requirement is a two way bridge between the acquisition software and external analysis platforms. When these bridges have been unavailable from microscope vendors excellent open-source tools such as Micropilot have attempted to fill this gap [4]. In our facility we use the Zeiss Open Application Development (OAD) framework as a two way bridge between Matlab and the Zen Blue acquisition software controlling the wide-field microscopes. With the setup we are able to send a variety of acquisition commands to Zen from Matlab and by using the Bio-formats library we are able to retrieve images, thus satisfying our two key requirements (fig. 1) [5].

Applications and Perspectives in Cancer Research

We use event driven microscopy for a variety of applications, here we briefly review several. Firstly, we have developed a search and find protocol for metaphase spreads (fig. 2). This protocol can save researchers large quantities of time manually searching slides for a few suitable spreads. Secondly, we acquire high resolution 3D fluorescence time-lapse movies of mitosis using bright-field, or phase-contrast, to detect the mitotic cells (fig. 3). Bright-field imaging requires much lower light levels than fluorescence hence less of the light-budget is wasted finding the mitotic cells.
We also use search and find protocols for the 3D confocal imaging of intestinal crypt stem cell colonies in thick, and large, tissue sections. The main challenge for this application is the varying topography of the sample. These protocols have facilitated the routine searching of several square centimeters of tissue, and the corresponding high resolution imaging of targeted events. Cancer Research UK’s Grand Challenge aims to produce 3D maps of tumors at cellular resolution which will yield vast quantities of information. We envisage that the availability of such reference maps will provide extensive opportunities for event driven microscopy. If we know what to look for then targeted imaging of interesting events in large cleared, or sectioned, tissue volumes can become commonplace.

[1] Fabrice de Chaumont, Stéphane Dallongeville, Nicolas Chenouard, Nicolas Hervé, Sorin Pop, Thomas Provoost, Vannary Meas-Yedid, Praveen Pankajakshan,  Timothée Lecomte, Yoann Le Montagner, Thibault Lagache, Alexandre Dufour & Jean-Christophe Olivo-Marin: Icy: an open bioimage informatics platform for extended reproducible research, Nature Methods 9:690–6 (2012) doi: 10.1038/nmeth.2075
[2] Johannes Schindelin, Ignacio Arganda-Carreras, Erwin Frise, Verena Kaynig, Mark Longair, Tobias Pietzsch, Stephan Preibisch, Curtis Rueden, Stephan Saalfeld, Benjamin Schmid, Jean-Yves Tinevez, Daniel James White, Volker Hartenstein, Kevin Eliceiri, Pavel Tomancak & Albert Cardona: Fiji: an open-source platform for biological-image analysis. Nature Methods 9:676–82 (2012) doi: 10.1038/nmeth.2019
[3] Anne E Carpenter, Thouis R Jones, Michael R Lamprecht, Colin Clarke, In Han Kang, Ola Friman, David A Guertin, Joo Han Chang, Robert A Lindquist, Jason Moffat, Polina Golland and David M Sabatini: CellProfiler: image analysis software for identifying and quantifying cell phenotypes. Genome Biology 7:R100. (2006) doi: 10.1186/gb-2006-7-10-r100
[4] Christian Conrad, Annelie Wünsche, Tze Heng Tan, Jutta Bulkescher, Frank Sieckmann, Fatima Verissimo, Arthur Edelstein, Thomas Walter, Urban Liebel, Rainer Pepperkok, and Jan Ellenberg: Micropilot: automation of fluorescence microscopy-based imaging for systems biology. Nature Methods;8:246–9 (2011) doi: 10.1038/nmeth.1558
[5] Melissa Linkert, Curtis T. Rueden, Chris Allan, Jean-Marie Burel, Will Moore, Andrew Patterson, Brian Loranger, Josh Moore, Carlos Neves, Donald MacDonald, Aleksandra Tarkowska, Caitlin Sticco, Emma Hill, Mike Rossner, Kevin W. Eliceiri, Jason R. Swedlow, Metadata matters: access to image data in the real world. The Journal of Cell Biology 189:777–82 (2010) doi: 10.1083/jcb.201004104

Jeremy Pike1, Patrice Mascalchi2, Joana Sarah Grah3, Stefanie Reichelt1

1 Cancer Research UK Cambridge Institute, University of Cambridge, Li Ka Shing Centre, Cambridge, UK
2 Bordeaux Imaging Centre, Université de Bordeaux, INSERM, Pôle d’imagerie photonique, Bordeaux, France
3 Department of Applied Mathematics and Theoretical Physics, Centre for Mathematical Sciences, University of Cambridge, Cambridge, UK

Jeremy Pike, PhD

Cancer Research UK Cambridge Institute
University of Cambridge
Li Ka Shing Centre
Cambridge, UK

Register now!

The latest information directly via newsletter.

To prevent automated spam submissions leave this field empty.