MitProfiler Validation study
Research type
Research Study
Full title
Validation of the MitProfiler algorithm for the detection and quantification of mitotic figures in haematoxylin and eosin stained tissue sections.
IRAS ID
342246
Contact name
David Snead
Contact email
Sponsor organisation
Histofy Limited
Duration of Study in the UK
0 years, 3 months, 31 days
Research summary
MitProfiler is a computer algorithm which uses machine learning, a form of artificial intelligence, to train computers to recognise mitotic figures in pathology samples (that is samples of human or animal tissue prepared for microscopic examination). Mitotic figures are seen in cells which are dividing. They are the chromosomes in the cells which can be seen preparing to divide and separating into the daughter cells as the cells actually divide. They are an important feature of normal tissue biology (e.g. in growing regenerating tissue) and diseases such as cancer (where cells grow and divide in an uncontrolled manner). Pathologists see and count mitotic figures routinely in their work. Having the computer do this task saves time, reduces fatigue in pathologists and should deliver more consistent results as observations between individual human pathologists varies to some extent. Variration in By getting human pathologists and the MitProfiler algorithm to count mitoses in the same samples, this study aims to measure how closely the MitProfiler algorithm performs in comparison to human pathologists. If the results are as good or better than human pathologists the data derived will be used to support the regulatory approval of the device through the Medicines and Healthcare Regulatory Authority.
REC name
London - Fulham Research Ethics Committee
REC reference
25/LO/0158
Date of REC Opinion
20 Feb 2025
REC opinion
Favourable Opinion