Resources - Application Notes

Detection and Automated Imaging of Regions of Interest (ROIs) when Performing Whole Slide Imaging (WSI)


Related Products: Cytation 7

May 15, 2020

ArrowRelated Sample File: Prostate tissue array H and E multiple ROIs

Author: Peter J. Brescia, Applications Scientist, Applications Department, BioTek Instruments, Inc., Winooski, VT


In today’s research and clinical environments, there is a movement towards Whole Slide Imaging (WSI) as a means to transform physical specimens mounted on traditional microscopy slides or various sample vessels into a digital medium. Once digitized, the data can then be easily stored as a shared resource for a variety of purposes including pathology, diagnostics, and as scientific research and educational tools. The digital format is amenable to analysis using both traditional human methods as well as increasingly powerful computational algorithms. Augmented microscopy allows for the development of streamlined methods to locate Regions of Interest (ROIs) on a slide, within a microplate well or within a given sample. Once ROIs are selected, automated image acquisition is performed using a range of available imaging methods and subsequent automated image analysis, as required.


Since the development of the first instruments to visualize microscopic objects by Antione Van Leeuwenhoek in the 17th century, efforts have been focused on improved methods to capture and analyze those objects1. While the first images consisted of hand drawings, some of which were quite elaborately detailed, in the latter half of the 20th century technological advancements allowed for digital imaging and subsequent computerized methods of analysis2. Sample variety spans the diverse biological and physical universe including tissue acquired in clinical settings for diagnosis, samples in basic research initiatives across disciplines and organizations and those suitable for educational and/or collaborative purposes3. Analyses include quantification of a range of observable parameters such as object counting and quantification of object size as well as more challenging metrics such as area, pixel intensity and subpopulation analysis of objects in multi-channel fluorescent images. Furthermore, digital images provide a means to standardize and automate analysis of some of the information present in a more precise, reproducible, and objective manner compared to traditional human analysis2.

Around the turn of the century the emerging focus on whole-slide imaging (WSI) of glass slides began to receive broader acceptance with improvements in rapid image acquisition and data management solutions4. Much of the driving force for this development was digitization of traditional histology slides that can be viewed via a computer monitor or mobile device for clinical pathology2. In particular, relieving the burden of routine image analysis of standard screening would significantly increase sample throughput3. However, the applications are very broad for such technology, having merits in academic research, pharmaceutical drug discovery and development programs as well as for biotechnology companies. The objective is to simplify the workflow by quickly identifying regions of interest (ROIs) using a low-magnification optical path followed by image acquisition of the selected region at a higher magnification.

The ROI feature can be useful for scanning a slide containing one or more tissue sections in color brightfield (CBF), for example H&E stained histological tissue sections or biopsy cores. The usefulness is extended by scanning fluorescently labeled tissue section(s) and imaging those ROIs using one or more channels at higher magnification such as for the assessment and quantification of biomarkers4. Additionally, more traditional imaging modes such as CBF, brightfield, and phase contrast can be combined with fluorescent channels resulting in rich data set. Provided here are two examples of the use of a combination upright and inverted automated image acquisition system for performing WSI employing ROIs where applicable.

Whole Slide Imaging (WSI).

Figure 1. Whole Slide Imaging (WSI). Upright camera and optics can quickly image a large number of sample mounted on a single glass slide at low magnification. Regions of interest (ROIs) are identified and subsequently imaged at higher magnification for analysis.

Materials and Methods


Cytation™ 7 Cell Imaging Multi-Mode Reader combines automated digital upright and inverted widefield microscopy with conventional multi-mode microplate reading in a unique, patented design. The inverted microscopy module provides sample visualization from 1.25x to 60x magnification in fluorescence, brightfield and color brightfield for a broad range of applications. The upright microscopy module with reflected light imaging enables even more applications such as ELISpot or fast slide scanning and ROI detection workflow. Cytation 7 includes continually variable bandpass monochromators for versatility and performance for general multi-mode plate reader applications. Temperature control to 45 °C, optional Peltier Cooling Module, gas control and shaking expand the applications for kinetic cell based assays. Cytation 7 is controlled by Gen5™ software, which combines ease-of-use with powerful processing and analysis capabilities.
Upright and inverted microscopes provide wide range of magnifications (1.25x-60x).
Figure 2. Cytation 7. Upright and inverted microscopes provide wide range of magnifications (1.25x-60x). The upright, low magnification optical path is used for Whole Slide Imaging (WSI). The inverted microscope is then used to image Regions of Interest (ROIs) at higher magnification using a variety of imaging methods (i.e., phase contrast, epifluorescence, color brightfield, etc.).


Stomach tissue, H&E stained (Carolina Biol. Supply Co., Burlington, NC, USA). Human Prostate Tissue MicroArray, H&E stained (Cancer) PN: NBP2-30169 (Novus Biologicals Europe, Abingdon, UK). WSI of the stomach tissue was performed at 2x magnification, 4x4 montage, using transmitted color brightfield via the upright digital microscope in the Cytation™ 7 to visualize and select ROIs. Three ROIs were selected and imaged at 10x magnification in color brightfield via the inverted WFOV digital microscope. WSI of the microarray was performed at 2x magnification, 3x5 montage, using transmitted color brightfield via the upright digital microscope to visualize and select ROIs. Fifty ROIs were selected and imaged at 10x magnification in color brightfield via the inverted WFOV microscope.


Human Prostate Tissue MicroArray

A microarray of prostate tissue cores containing 40 adenocarcinoma samples, with histologic grades (Gleason scores) ranging from 6-10, and 9 matching normal tissue samples, from patients between the ages of 44 and 75 years old, were H&E stained for imaging. Sample cores are 2.0 mm in diameter with a section thickness of 4 μm arranged in a 5x10 matrix. Initial WSI was performed using the parameters listed in Table 1 generating a single large, low-resolution image of an area slightly larger than that containing the tissue cores. ROIs were selected for each core to minimize imaging of the background area. The ROIs were subsequently imaged at 10x magnification using the parameters listed in Table 1 resulting in 50 unique sets of tiled images (3x3 montage) for each sample core; 49 prostate tissue samples and one carbon location marker. The images were processed into a single image using automated image stitching resulting in 50 individual images for future analysis and archiving (Figure 3).
Cytation 7 imaging settings.
Table 1. Cytation 7 imaging settings.
Human Prostate Tissue MicroArray.
Figure 3. Human Prostate Tissue MicroArray. (A.) Low-resolution image of microarray, (B.) representative image of a 3x3 stitched montage of a prostate adenocarcinoma using WFOV camera at 10x magnification and (C.) representative image of normal matched prostate tissue as describe above.

Mammalian Stomach Tissue: Cardiac, Fundic and Pyloric Regions

Three sample tissue sections representing the cardiac, fundic and pyloric regions of the stomach with a section thickness of ~4 μm arranged under a 18 mm, circular coverslip were imaged. Initial WSI was performed using the parameters listed in Table 1 generating a single large, low-resolution image of an area slightly larger than the coverslip. ROIs for each tissue section were selected to minimize background area and imaged at higher resolution using the parameters listed in Table 1 resulting in three unique sets of tiled images (3x6, 2x4, and 3x7 montages) for cardiac, fundic and pyloric regions, respectively. Tiled images were stitched into single images using automated image stitching resulting in three individual images for future analysis and archiving (Figure 4).  
Mammalian Stomach: Cardiac, fundic and pyloric regions.
Figure 4. Mammalian Stomach: Cardiac, fundic and pyloric regions. (A.) Low-resolution image of tissue samples, (B.) selection of ROIs, (C.) stitched image of ROI at 10x magnification and (D.) zoomed area representative of high-resolution stitched image.


The ability to capture images digitally has led to a paradigm shift resulting in simplified workflows and sharing for diagnostics, education, and basic scientific research. Automated imaging solutions have further expanded the capabilities by offering rapid whole slide imaging (WSI) or other various vessel types to determine regions of interest (ROIs) for subsequent higher resolution imaging and analysis. Additionally, the inclusion of multiple imaging modes such as color-brightfield, phase constrast and epifluorescence microscopy greatly expands the breadth of possible applications. Two examples shown here - imaging of prostate tissue cores and stomach tissue sections in color-brightfield - show the ability to rapidly image entire vessel surfaces at low resolution, select only those regions of interest, and capture highresolution images of the ROIs using automated imaging methods.

This results in a powerful system, that when coupled with image-based computational analysis tools, provides a viable solution to increase sample throughput.


  1. Meijer GA, Beliën JA, van Diest PJ, Baak JP. Origins of… image analysis in clinical pathology. J Clin Pathol. 1997;50:365–70.
  2. Aeffner, Famke; Zarella, Mark D.; Buchbinder, Nathan; Bui, Marilyn M.; Goodman, Matthew R.; Hartman, Douglas J. et al. (2019): Introduction to Digital Image Analysis in Whole-slide Imaging: A White Paper from the Digital Pathology Association. In Journal of Pathology Informatics 10. DOI: 10.4103/jpi.jpi_82_18.
  3. Gurcan, Metin N.; Boucheron, Laura E.; Can, Ali; Madabhushi, Anant; Rajpoot, Nasir M.; Yener, B. (2009): Histopathological image analysis: a review. In IEEE reviews in biomedical engineering 2, pp. 147–171. DOI: 10.1109/RBME.2009.2034865.
  4. Zarella, Mark D.; Douglas Bowman; Famke Aeffner; Navid Farahani; Albert Xthona; Syeda Fatima Absar et al. (2019): A Practical Guide to Whole Slide Imaging. A White Paper From the Digital Pathology Association. In Arch Pathol Lab Med 143, pp. 222–234.