The overall goal of this project is to employ deep learning techniques to improve treatment stratification for patients with colon cancer. By integrating both pathology and radiology data, as well as clinical data, the aim is to develop predictive models that can optimize treatment decisions and improve patient outcomes.

---

Objective

The ultimate goal of this project is to design an artificial intelligence-based model that enables an accurate assessment of the optimal treatment strategy for an individual patient with non-metastatic colorectal cancer, at the time of diagnosis—that is, before any treatment has begun. On a secure website, the patient, together with his or her physician, will be able to upload the digitized images of the CT scan and tissue biopsy. The platform will then provide an estimate of the risk of recurrence, which can serve as the basis for the treatment plan.

Methodology

1. Collect a retrospective database of localised colon cancer patients.
2. Develop a deep learning algorithm to predict recurrence based on diagnostic biopsy data.
3. Develop a deep learning algorithm to predict recurrence based on diagnostic CT imaging data.
4. Develop a multi-modal deep learning algorithm to predict recurrence based on diagnostic biopsy and imaging data.

Impact and future directions

The clinical finality of this project is an open access web-based platform on which radiological and pathological images can be uploaded at the time of diagnosis, resulting in an outcome prediction for an individual patient. Using this platform, patients eligible for neoadjuvant chemotherapy can be identified while minimising the risk of overtreatment.

General info & contact

Keywords (#): Colon cancer, treatment decision, AI-driven, multi-modal, software tool

RADar project research lead: Dr. Nathalie Mertens

RADar project researchers: Prof. Dr. Peter De Jaeger

Principal investigator: MD Sofie De Meulder

Timeline: 2024-2028

Status: Ongoing

Publications / presentations: –

Partners: UZ Leuven

Funding: Roche, MSD, Amgen, Nordic, Stichting Tegen Kanker, Kom Op Tegen Kanker