VALIDATING AI IN CLASSIFYING CANCER IN REAL-TIME SURGERY

Projektinformationen

CLASSICA

ID Finanzhilfevereinbarung: 101057321

DOI

10.3030/101057321

Projekt abgeschlossen

EK-Unterschriftsdatum 19 Mai 2022

Startdatum 1 Mai 2022

Enddatum 30 April 2026

Finanziert unter

• Health

Gesamtkosten

€ 5 978 718,74

EU-Beitrag

€ 5 978 716,00

Koordiniert durch

UNIVERSITY COLLEGE DUBLIN, NATIONAL UNIVERSITY OF IRELAND, DUBLIN
Ireland

Periodic Reporting for period 2 - CLASSICA (VALIDATING AI IN CLASSIFYING CANCER IN REAL-TIME SURGERY)

Berichtszeitraum: 2023-11-01 bis 2025-04-30

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

CLASSICA is deploying cancer classification AI technology in the operating theatre. We will deliver and clinically validate an AI-based decision support technology, which allows rapid identification of the presence and distribution of cancer within tumours identified via endoscopy, and in particular the rectum. Validation will be carried out across several hospitals, surgical teams, and countries.
Colorectal cancer (CRC) is the third most common cancer type globally and the second leading cause of cancer death, with nearly one million deaths annually. The project addresses a key clinical challenge: the treatment of significant rectal polyps, especially those that are sessile and larger than 2 cm. While smaller polyps are typically managed through routine endoscopic polypectomy and advanced cancers through established paradigms, these significant polyps may require advanced local excision, such as transanal endoscopic microsurgery (TEM) resection. CLASSICA addresses the unreliability of current methods for characterisation of these polyps, such as biopsies (unreliable for 46%-59% of such lesions) and radiological assessments (tend to overestimate 50% of these lesions). This is problematic because accurate pre-operative characterisation can significantly influence the choice of endoscopic technique and the dissection plane.
Our technology in CLASSICA leverages the rapid analysis of tissue fluorescence dynamics to distinguish between cancerous and non-cancerous tissues during surgeries. This approach is based on the variance in fluorescence intensity of tissues dyed with fluorescent Indocyanine Green (ICG) and illuminated by near-infrared (NIR) light. Our classification software can provide a prediction on the likelihood of the presence of cancer within the whole tumour and also identify the area of the most abnormal flow signal, which can then be used for targeted biopsy. This is done in real-time in the operating room (OR).
By integrating this technology into surgical practice, CLASSICA not only contributes to the field of medical technology but will also help ensure the patient proceeds to the optimum treatment pathway and reduce the need for further surgery. Importantly, too, the consortium contains work packages devoted to surgeon and public education and information, and also concerning the ethics and legal (including liability) aspects of AI-driven surgical interventions. CLASSICA represents a significant stride in enhancing surgical care for CRC, combining technological innovation with the expertise of surgeons to provide a more precise, reliable, and effective treatment approach.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

The CLASSICA project is making substantial progress in the development and clinical validation of our AI-assisted surgical decision-support system. CLASSICA is now active across seven clinical sites in Ireland, Belgium, Austria, Italy, the Netherlands, and Greece. Clinical validation is structured around three studies that collectively assess the system in diverse surgical contexts.
Study 1 focuses on generalisability and usability and continues to collect intra-operative video data to develop and evaluate the system’s performance in varied clinical environments. To date, over 150 procedures have been annotated and analysed, enabling cross-site comparisons and iterative improvement of the AI models.
Study 2, an interventional trial investigating AI-guided biopsy in the OR, was launched at the Mater Misericordiae University Hospital in Dublin during this reporting period. The first patients have been successfully recruited, and the CLASSICA software has been deployed for use in live procedures.
Study 3, which will evaluate AI-assisted resection, remains in the preparatory phase, with updates to follow in the next reporting period.
Over the course of the project, the CLASSICA AI pipeline has been refined to improve tissue tracking, image enhancement, and fluorescence-based classification.
To ensure technical standardisation and support clinical adoption, comprehensive protocols and standard operating procedures were finalised, covering patient selection, ICG fluorescence administration, intra-operative video capture, and the use of CLASSICA software systems. These support safe and consistent implementation across sites. A robust data management framework has been established, with safeguards at all stages to ensure compliance with GDPR and MDR requirements, including trial monitoring and quality checks at trial sites.
Training has been a key focus, with instructional materials and site-specific onboarding ensuring that clinical users are prepared to use the CLASSICA system in the OR. Resources include educational videos and professional masterclasses for surgeons and OR staff. These videos demonstrate the standardised clinical method for capturing surgical footage and best use of the CLASSICA software.
The groundwork for enhancing clinical guidelines, particularly for TEM procedures, has been laid. This includes the formation of the EAES Task Force and conducting legal, ethical and regulatory framework studies to shape guidelines that are practically applicable and widely accepted.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

The following are key areas where CLASSICA will deliver results beyond the state of the art:
- CLASSICA introduces cutting-edge AI algorithms for real-time tissue classification.
- CLASSICA will empower surgeons to make more informed decisions, potentially reducing the need for additional surgeries and improving patient outcomes.
- Our validation studies across 8 hospital sites will demonstrate the software’s effectiveness and adaptability in different clinical environments.
- CLASSICA is addressing the evolving regulatory and ethical landscape, and biases and liability concerns related to AI-assisted surgery.
In terms of communication and exploitation, CLASSICA has laid a strong foundation with extensive public engagement. To support long-term uptake, CLASSICA convened an expert-stakeholder congress during the Life Sciences AI Summit 2025, bringing together clinicians, industry, regulators, and policymakers. IP management is underway, with the first Exploitation Workshop guiding our commercialisation plans. Our communication strategy has yielded substantial engagement, evidenced by a strong online presence. Key dissemination achievements include publishing 23 journal articles, 5 book chapters, and notable participation in major conferences.

CLASSICA logo