Histology Imaging and Analysis

Whole slide scanning is an automated technique where a tissue section on a slide is detected, a series of images are scanned across the whole sample at high magnification and then montaged so describe the tissue sample. When imaging and mathematically modeling the whole or part of the tissue there are concerns that need to be addressed

  • Standardisation of the tissue during collection, preservation and preparation
  • The time taken to image and assess the tissue
  • Multiple labeling of tissue
  • Image analysis

image of the gut

The above image shows part of a data set from primary gut tissue in a study linking obesity, Insulin-like growth factors and cancer. The image is a small subset of the total, only containing ten images from a total dataset of 1,000 fields of view. To aid transmission over the internet the image has been reduced in size by a factor of five.

Image: D.Roberts and A. Renehan, Colorectal Cancer Focus group


Tissue preservation and preparation

Please visit Garry Ashton’s Histology Service pages for more information [link]


To fully describe a ‘standard’ histology slide at x400 total magnification, around 10,000 - 80,000 fields of view needs to be examined. By automating the process of data collection batches of slides can be imaged, where is it possible to digitize 300 slides overnight. Since 2007 this technique has been utilised in the laboratory which has produced around 100TB of data per year.

Two methods are employed within the Institute for the visualisation of data

  • using filters and a camera to scan the tissue as a whole
  • spectrophotometrically imaging across the visible spectrum and sampling the tissue. Used in the identification and assessment of rare cells within the tissue.

an example of whole slide scanning with tumour tissue

Using the whole slide scanning tumour tissue has been visualised at x400 total magnification. The images above demonstrate tissue scanning of samples to visualise CD44 at a range of magnifications from the whole tissue, and then zooming in by x10, x20 and x40. In the lab, CD44 is used as a as cell surface marker for some breast and prostate cancer stem cells. CD44 is a glycoprotein involved in cell–cell interactions, cell adhesion and migration.  

Image: Francesca Trapani & Caroline Dive, Clinical and Experimental Pharmacology

Multiple labeling of tissue

For the researcher to be able to put multiple labels onto tissue allows the quantification of relationships between molecules. In the field of personalised medicine being able to mathematically model molecular interactions is vital and often difficult to carry out successfully.

With methods of imaging that require filters and a camera there is a limit to the number of labels that can be discerned for analysis. This is due to the wide emission spectral of fluorescent labels, wide spectral range of filters and the overlapping spectral properties of labels. Using these methods five labels can be separated with fluorescent labeling and 2-3 with immunohistochemistry.

The other difficulty with tissue imaging is auto-fluorescence. Many unlabeled tissues exhibit spectral properties which interfere with multiple labeling and analysis. Common molecules such as NADH and flavins all exhibit spectral properties down at the green part of the spectrum thus labels such as Alexa fluor 488, and 512, green fluorescent protein, yellow fluorescent protein are almost impossible to separate from the ‘sea of fluorescence’ visible across the tissue due to auto-fluorescence. Within the laboratory certain tissues such as bladder and kidney has extraordinary levels of auto-fluorescence and so study of some cancers are difficult. Within the laboratory the Perkin-Elmer Vectra has been installed for multiple label imaging and spectrally removing autofluorescence thus allowing the imaging of 'difficult' tissues. Working alongside the histology service we are putting together procedures and routines for the multi-labeling of tissue beyond the limits of the number of labels that can be applied.

Analysis of tissue

Traditionally methods of separating labels and structures via thresholding have been used to separate and mathematically describe data, however on close examination of the tissue immunohistochemistry produces labeling over a range of intensities. This variation in labeling intensity makes standard intensity based image analysis difficult. The other difficulty with image analysis of immunohistochemicaly labeled hsitology is that the tissue is on a white background. Separation of labels is difficult as all of the colours that you could potentially separate and numerate from the tissue are also present in the background.

Two methods are utilised within the institute for the analysis and mathematically modeling of tissue

  • Definiens (with tissue studio and image miner): contrast based analysis of the whole tissue.
  • PerkinElmer inForm: the analysis of cells or tissue when imaged via a spectrophotometer.


cell nuclei and membranes via definiens tissue studio

Using the tissue detailed as above, Definiens Tissue Studio was used to classify tumour (orange) and stroma (blue) in tumour tissues stained for CD44.  Image 1 the whole tissue, image 2 a subset of the total section; Image 3 colour mask applied during training of the two data subsets, and Image 4 separation of tissue and stroma with the mask removed. The analysis is carried out across the whole tissue, a subset is demonstrated above.

Image: Francesca Trapani & Caroline Dive, Clinical and Experimental Pharmacology


cell nuclei and membranes via definiens tissue studio

On the tissue outlined above, Definiens Tissue Studio algorithms were used to identify cell nuclei and membranes (image 3) and subsequently classification of all cells in the tissue section in low (yellow), medium (orange) and high (red) based on CD44 staining intensity (image4). The analysis is carried out accross the whole tissue, a subset is demonstrated above.  

Image: Francesca Trapani & Caroline Dive, Clinical and Experimental Pharmacology

Institute equipment

The facility has been whole slide scanning and analysing the histological data from immuno-histochemistry and fluorescence since 2007. At present there are three systems, for data collection:

  • PerkinElmer Vectra: imaging across the visible spectrum (420-750nm) with a 200 slide autoloader
  • Leica SCH400: whole slide imaging of immunohistochemistry labeled tissue with a 300 slide autoloader
  • 3D-Histech Mirax: whole slide imaging of immunohistochemistry or fluorescently labeled tissue with a 150 slide autoloader

For the analysis of data several software packages are utilised

  • PerkinElmer inForm 
  • Definiens
  • Definiens Tissue Studio
  • Definiens Image Miner
  • Voloom, for the alignment of serial sections to 3D volumes

The facility offers assistance in the collection of data, visualisation and the analysis of the data, in addition the Histology Service offers assistance in planning labeling regimes, and has equipment to ensure standardisation of sample preparation and labeling and the Scientific Computing facility offers a fully data storage, backup and archival service.