Staged AI analysis of tissue slides and scans to predict complications after colorectal cancer surgery
Official title: Staged Unimodal-to-Multimodal AI Analysis of Histopathology, CT/MRI, and Multiplex Tissue Imaging for Perioperative Risk Prediction in Colorectal Cancer (KIA-Korekt)
Run by Medizinische Hochschule Brandenburg Theodor Fontane · NCT07537491 · Data updated 2026-11-12
This project tries to use AI on tissue slides, CT/MRI scans, and advanced tissue imaging to see if it can predict who will have complications after colorectal cancer surgery.
Quick facts
| Study type | Observational |
|---|---|
| Ages | 18 Years and up |
| Sex | All |
| Enrollment | 910 (estimated) |
| Sponsor | Medizinische Hochschule Brandenburg Theodor Fontane Academic / other |
| Locations | 1 site (Brandenburg an der Havel) |
| Status | Recruiting |
| Started | January 1, 2011 |
| Main results expected | December 31, 2027 |
| Study ends | June 30, 2028 |
| Registry record updated | April 17, 2026 |
| Trial ID | NCT07537491 on ClinicalTrials.gov |
Is this trial a fit for you?
✓Likely a good fit
Adults (≥18) with histologically confirmed colorectal adenocarcinoma undergoing surgical resection who have available H&E whole-slide images (and preoperative imaging where available) are the intended participants.
✕Probably not a fit
Patients who are not undergoing surgery, who lack usable H&E slides or adequate imaging, or who have non-adenocarcinoma colorectal cancers are unlikely to benefit from the model.
A plain-English reading of the study's rules. Only the study team can confirm whether you qualify.
What this trial studies
KIA-Korekt combines retrospective imaging and pathology data with a planned prospective validation cohort to develop AI models for predicting perioperative complications in colorectal cancer. The retrospective set includes about 750 patients treated between 2011 and 2021. The approach integrates three imaging modalities — H&E whole-slide histopathology, preoperative CT/MRI radiomics, and multiplex tissue imaging (mIHC and IMC) — and applies convolutional neural networks and attention-based multiple instance learning alongside automated segmentation and radiomics pipelines. A multi-metric quality-control process is used to harmonize data and train models to predict short-term outcomes such as anastomotic leakage, wound infection, thromboembolism, and in-hospital mortality.
Why it matters
Potential benefit: If successful, the tool could help surgeons and care teams identify patients at higher risk of complications so they can tailor perioperative planning and monitoring to reduce harm.
How similar studies have gone: Prior AI and radiomics studies have shown promise for prognostic and diagnostic tasks in oncology, but combining multimodal imaging specifically to predict perioperative complications is relatively novel and not yet proven in large clinical cohorts.
Who can join: the full rules
Show full inclusion / exclusion criteria
Inclusion Criteria: * Adult patients (≥18 years) * Histologically confirmed colorectal adenocarcinoma * Undergoing surgical resection (curative or palliative intent) * Availability of H\&E-stained whole-slide images (WSIs) from the primary tumour Exclusion Criteria: * Patients not undergoing surgical treatment * Missing H\&E-stained tissue slides of the primary tumour * Histopathological material of insufficient quality for analysis
Where it's running
-
University Hospital Brandenburg an der Havel, Brandenburg an der Havel, Germany (Single-center)
Brandenburg an der Havel, Germany Recruiting
Study contacts
- Study coordinator: Melissa Horner, MSc
- Email: melissa.schadl@mhb-fontane.de
- Phone: +4915207809673
How to take part
- Review the eligibility criteria above with your treating physician.
- Check the official listing on ClinicalTrials.gov for the latest contact details and recruitment status.
- Contact the study coordinator or principal investigator to ask for pre-screening. Pre-screening is free and never obligates you to enroll.