A microscope with a green light ©

All rights reserved to the British Council


The process of examining stained tissue samples to reveal their biological secrets under the microscope is called histology. Student histopathologists can use a histology atlas to develop their skills in reading these intricate maps of cellular structures. Dr Raza Ali, is part of a team at Cambridge University and Cancer Research UK, that has taken this one step further by adapting computer software, originally designed for analysing distant galaxies, to the task of checking digitised cancer tissue samples.  


Ali explains how large clinical studies of patients involves characterising the digitised images of tumour samples by a histopathologist. Since many clinical studies involve thousands of patients, this manual quantification slows down the progress of the studies. 

Survey astronomers encounter an analogous problem: they need to repeatedly and accurately analyse thousands of images of the night sky. Ali says, ‘we have exploited this overlap in image analysis to develop automated methods in order to overcome the bottleneck of manual assessment.’

The painstaking process involves the coordinated placing of hundreds of tumour samples onto small glass slides. The samples are chemically treated so that the presence of particular proteins results in brown-staining. Then the slides are digitised and the digital images (one per tumour) are sent to the Institute of Astronomy. There the images are modified into an astronomy-format including inversion of the colours so that that dark colours become bright – like stars in the night sky. Depending on the protein, images are processed to quantify the amount of relevant brown stain in each tumour sample.


The team intends to use many more samples to determine the reliability of the method and to develop sophisticated analyses to evaluate more features of the images. Ali explains how the challenge of dealing with thousands of samples can be overcome by designing a robust structure to deal with the scale of the endeavour.  A more difficult problem is the sheer diversity of biological samples, which means it can be difficult to anticipate and adjust for all permutations.

This novel approach has led to an automated system for checking tumour samples, which Ali hopes will lead to faster assessment of tumour development in biopsies such as for breast cancer and support better targeted treatments. 


Why not do a language activity based on this cubed story, Automated Cancer Diagnosis?