Background
Mitotic density is an important prognostic factor in diverse tumors, which is typically assessed visually by trained pathologists using microscopic images. It has been shown that reproducibility of mitotic density may be suboptimal and that counting mitoses is tedious and time consuming. In previous we investigated the possibility to apply artificial intelligence (AI) to support pathologists in this task, and we showed a potential decrease in time consumption as well as observer variability. However, previous studies were mostly small scale, and no translation of AI into clinical practice has yet been performed.
Aim
The EFRO-funded AIRAT project aims to rigorously validate AI as a support for mitotic counting in breast cancer and implement the AI models in clinical practice. The validation study is set up to meet the strict requirements of in vitro diagnostic regulation (IVDR) certification, involving >25 (internationally recruited) pathologists and >200 breast cancer cases from multiple sources, as well as a strict statistical analysis plan.
Funding
The AIRAT project is funded by the 'Europees Fonds voor Regionale Ontwikkeling' (EFRO) Oost-Nederland under grant number OOST-00002.
