Endoscopic Computer-aided Diagnostics based on convolution neural network in detection of early phase adenoma
Description: The COMRADE project aims to create a prototype of the computer-aided diagnosis system based on the operation of convolutional neural networks(CNN) that would help detect early malignant lesions (adenomas) in the large intestine.The fundamental problem described in the project is the fact that human eyes do not allow on rapid and efficient detection of small flat lesions withinthe intestinal lumen. This problem is mainly related to the commonly used white light endoscopy in the world, which usually has no magnification. Thepossibilities of early detection of adenomas, in this case, are therefore dependent on two factors: the level of experience of the operator and theequipment at his disposal. To improve the quality of the results and reduce the margin of error of misdiagnosis (exclusion of false-negative results), it isnecessary to increase the accuracy and precision of the detection of adenomas. Such diagnostics is crucial for patients aged 50+ who are in the riskgroup of developing colorectal cancer and all patients with inflammatory bowel disease. The adenoma detection rate (ADR) is a widely acceptedinternational benchmark on the quality of screening colonoscopy. A physician’s adenoma detection rate is the proportion of individuals undergoing acomplete screening colonoscopy who have one or more adenomas or polyps detected.The project estimates that ADR of a good endoscopist, calculated at ≥25% could be increased to around 80% by utilizing the CNN model implementedin the COMRADE system. Such high effectiveness will depend on the quality of the public database of colonoscopic photos and films, which will becreated for the project purpose. The project team estimates that there will be cooperation in the creation of a public database of colonoscopic photosand films with leading clinical units in Poland
Funder: Narodowe Centrum Badań i Rozwoju
Amount: PLN 1,498,000
Funding type: research
Funder scheme: LIDER
Start date: 2022-03-01
End date: 2025-12-31
Investigator: Marcin Talar
Institution: Uniwersytet Medyczny w Łodzi
Funded outputs count: 0
Funder award ID: LIDER/50/0262/L-12/20/NCBR/2021
OpenAlex ID: G6481975733