What is High Content Imaging?

Delving into the specifics: high content imaging is generally described as a highly automated three-step process used to dramatically increase imaging throughput and quantification. These three steps are:

1. Image acquisition
2. Image processing
3. Image analysis

Each step is associated with high degrees of computer-driven automation, allowing users to accelerate repetitive operations and eliminate human bias from the process. In this fashion, Tera-bites of imaged data can be acquired, processed and analysed within minutes, with minimal human operation. Such data often covers a variety of biological assays that, using traditional manual microscopy & analysis tools, would take between weeks and months to complete.

The image acquisition technology behind HCI seeks to rapidly place the focal plane of an objective lens within a biological sample to create a sharp image automatically and without human intervention. The requirements include ultra-fast autofocusing and sample detection, accurate and repeatable precision-motion modules, and enhanced acquisition speeds through multiple cameras, polychromatic illumination and/or spinning disc pinholes. When put together, these systems enable high content imaging platforms to extract detailed, multi-parametric data at the single-cell and subcellular level.

Once acquired, the images may need further processing that must be done in an automated fashion, and ideally in batch.  Below is a list of common image processing operations.

• Image stitching: joining several, adjacent fields of view together into a single image that visualizes a large, continuous area of the sample.
• Z-stack processing: reducing the 3D volume visualized to a single 2D image.
  • Selection of specific z-planes, either through user choice or analysis to identify those having the sharpest focus, is a simple operation to obtain a representative 2D image.
  • Intensity projections generate a new, representative 2D image by:
    • First, extracting the intensity information contained in the sample volume covered by each pixel area.
    • Second, performing a statistical operation, such as average or maximal intensity calculation.
    • Each pixel in the new image then records the output of this statistic through the volume.
  • Fluorescence deconvolution: improve contrast in images, most commonly when using widefield microscopes.
    • This mathematical operation estimates the contribution of out-of-focus fluorescence within each image and removes it.
    • The output image has higher contrast and is sometimes referred to as a digital confocal image.

Several of these operations can be performed together, such as deconvolution followed by intensity projection and stitching.  Batch processing to perform the operations over several datasets saves substantial time, as researchers do not need to load datasets one at a time for processing. 

Once images are processed, automated image analysis extracts the data of interest from the acquired images.  As described in our post , this approach is the heart of High Content Analysis (HCA).  It begins with segmentation to identify objects of interest, then extraction of metrics for each object, such as intensity and morphology information.  The power of HCA is that once parameters for segmentation are identified and defined, they can be applied to several data sets without bias. 

Analysis of processed images can sometimes be preferable over raw images, such as instances when studying interactions involving many cells, such as confluency, colonies of cells, intercellular signalling or infection studies (e.g., viral plaque formation), or tissue slices.  In these cases, stitched images allow for segmentation of large area objects that may extend beyond the edges of any single, raw image acquired.  Intensity projections can expedite identifying cells in Z-stacks and avoid potential double-counting of cells if 3D volumetric analysis is not performed.   Finally, deconvolution can assist with identification of small, intracellular structures such as foci, granules or organelles, where background fluorescence can impede reliable segmentation.