HIstopathology-based predictions by Multi-modal data Integration for COlorectal cancer diagnostics (HIMICO)
- Project Identification
- Project Period
- 6/2022 - 12/2022
- Investor / Pogramme / Project type
- Masaryk University
- MU Faculty or unit
- Faculty of Science
HIMICO brings together the top European researchers with extensive prior biological and pathological knowledge in the field of
colorectal cancer (CRC), together with all relevant AI image analyzers that have already proven track record in this disease, while using
the state-of-the-art technologies.
We will integrate multi-modal data on colorectal tumors to design prognostic models by top-down, bottom-up and cell community
approaches. We use a combination of a deeply phenotyped dataset with large clinically relevant datasets for >12,000 tumors. For all
H&E images with patient metadata including age, gender, disease information allowing to diversify patient information across
training and test sets which will result in robust and reproducible models, while the combination with the multimodal dataset will
ensure biological interpretation. HIMICO thus improves the current state-of-the-art in CRC digital pathology and biological drivers of
metastasis by building biologically interpretable prognostic models for predicting relapse-free survival (RFS). As opposed to black box
machine learning approaches, these approaches will develop transparent and accurate computational models allowing extraction of
underlying biology that can suggest novel therapeutic nodes.
Our combined specific expertise is currently not present anywhere worldwide, making this a highly competitive project with a synergistic team and 2 SME guaranteeing effective translation.
Gained technological expertise will lower the health care burden for CRC patients in the EU and worldwide. It can provide a better and
more cost-efficient solution, tailored to the needs of individual CRC patients through technological integration of multimodal
datasets. With our approach we will implement quality and risk management systems to ensure compliance with the European
approach to artificial intelligence and to minimize risks for scientists, patients, and clinicians.