The list of award winners in 12^th Leader programme by National Centre for Research and Development has been revealed and we are proud to report that Marcin Talar from Department of Biochemistry MUL was one of the winners. His project, Computer-assisted endoscopic diagnostics based on convolutional neural networks to detect early neoplastic lesions (adenomas) in the colon was ranked 12 on the list of a total of 56 applications. The project total value amounts to nearly PLN 1,5 million.

The aim of COMRADE (COMputeR-aided Diagnostics of ADEnoma) project is to design a computer-assisted system of medical diagnostics based on convolutional neural networks (CNN), which will allow for the detection of early neoplastic lesions (adenomas) during colonoscopy procedures. Currently on medical market, we observe a considerable demand for artificial Intelligence tools which could facilitate the research conducted and health care provided by doctors in hospital and clinical units.

The successful project involves an innovative solution to create a detection algorithm so that, apart from neoplastic lesions detection, it could provide an automatic unified record of the endoscopy of gastrointestinal tract. The task will be carried out on the basis of the developed MST (Minimal Standard Terminology) nomenclature that is used in descriptions of endoscopic imaging and segmentation of characteristic areas of the large intestine which will jointly create a useful diagnostic map, says Marcin Talar MD.

As part of LEADER programme, our University server room shall be expanded by adding new computing units using graphics processors which will be capable of quick analyses related to the processing of data from medical images. The project also involves a formation of an interdisciplinary team of researchers including MUL gastroenterologists and engineer-programmers from Technical University of Lodz. Of note, the awarded project has been approved by leading providers of imaging and data analysis services.

Medical University of Lodz, which supervises 3 university hospitals, is an excellent source that provides a large number of photos and film sequences on lower gastrointestinal endoscopy. Moreover, the project assumes a collaboration with top national clinical centres which perform colonoscopy in order to create and keep a nationwide, public, anonymised database of colonoscopy films and images.

We trust that the outcomes of this interdisciplinary projects from the border of medicine and bioinformatics which uses convolutional neural networks will have a chance to revolutionise colonoscopy research in Poland and worldwide, to significantly increase its effectiveness and, last but not least, to contribute to faster and more precise diagnostics in patients, adds the project author.

Congratulations and we wish further success in scientific career!