Insights from the CLASSICA Study
The CLASSICA project has published a new “Knowledge Unwrapped” article that summarises the scientific paper “CLASSICA: Validating artificial intelligence in classifying cancer in real time during surgery,” originally released in Colorectal Disease (Volume 25, Issue 12, 2023). The publication, created by a multidisciplinary group of clinicians and researchers from across Europe, includes contributions from Jernej Cucek in Jan Rojc (Arctur).
The paper outlines the study protocol used to examine how the CLASSICA AI system can support surgeons by classifying colorectal lesions in real time during an operation. The work explains how AI-based assessment is compared with standard diagnostic methods, such as biopsy and MRI, using video recordings from colorectal procedures to evaluate accuracy and clinical applicability.
Why this research matters
Colorectal cancer remains a significant global health burden, and precise assessment during surgery is essential for guiding treatment. Traditional diagnostic tools can be limited in providing immediate clarity. CLASSICA investigates how AI could improve real-time decision-making, reduce misdiagnoses, and help avoid unnecessary or overly aggressive interventions.
Methods described in the publication
The study examines performance of the CLASSICA AI model by analysing surgical video footage and comparing its outputs with conventional diagnostics. The protocol focuses on validating reliability, ensuring that AI-generated classifications can support surgeons safely and effectively.
Potential impact
The work highlights how AI integration could:
- improve diagnostic accuracy during colorectal cancer surgery,
- reduce unnecessary procedures and overtreatment,
- strengthen surgical decision-making, and
- support the introduction of AI tools into clinical practice.
Next steps
The following research phases will compare AI-guided targeted biopsies—or, where appropriate, AI-only lesion characterisation—with standard biopsy approaches. Studies will also assess AI-supported local excision versus traditional excision methods, focusing on margin clearance and recurrence.
The full paper is available through the CLASSICA Zenodo community, together with other project publications.