New method to identify cell types using Artificial Intelligence

Researchers from Swansea University’s College of Engineering have developed a new method to identify different cell types – such as cancer cells – by training computers to detect them using Artificial Intelligence (AI) algorithms. The method uses a similar approach to identification as face or fingerprint recognition software.

The research is an international collaboration with the Broad Institute of MIT and Harvard in Cambridge, Massachusetts, USA; Helmholtz Zentrum Munchen in Munich, Germany; The Francis Crick Institute in London; and Newcastle Upon Tyne University.

The group’s paper, entitled Label-free cell cycle analysis for high-throughput imaging flow cytometry’, is published in the leading life sciences journal Nature Communications today (Thursday, January 7).

AI calls image

 

  

 

 

 

 

One of the paper’s authors, Professor Paul Rees (pictured below) of Swansea University’s College of Engineering, said: “To identify different types of cells, e.g. cancer cells, within a health cell population, scientists usually have to use special fluorescent stains that bind to components of the cell to allow detection using microscopy.  

Professor Paul Rees“Unfortunately these stains alter the cell’s behaviour and modify the system being investigated.

‌‌“The new method we have developed avoids the use of these stains using AI machine learning algorithms.  The researchers train the algorithm to recognise the specific cell of interest by giving examples of the cell to be identified.

‌“After learning what the cells look like, the computer algorithms can then identify the target cells in a population of previously unseen cells.”

The new method is so accurate it is also able to determine the position of the cell within its life cycle.

“Most anti-cancer treatments act specifically on cells at a certain point within their life cycle and therefore it is highly desirable to determine the age of cells in a population, without perturbing them with stains,” added Professor Rees.

“Computer-based classification of cells with algorithms opens up a whole new perspective that could also be used for entirely different research questions, not only for cell cycle analysis,” added Professor Dr Dr Fabian Theis of the Helmholtz Zentrum Munchen.

Professor Paul Rees’ work was supported by the Engineering and Physical Sciences Research Council’s (EPSRC) International Collaboration Sabbatical scheme.

The work at Swansea University, which also involved Professor Huw Summers of the College of Engineering, was supported by the Biotechnology and Biological Sciences Research Council (BBSRC) and the National Science Foundation (NSF) in the USA.


The full paper – Blasi, T. et al. Label-free cell cycle analysis for high-throughput imaging flow cytometry. Nat. Commun. 6:10256 doi: 10.1038/ncomms10256 (2015) – can be viewed here.