Digital pathology and detection of lymph node metastases
Research type
Research Study
Full title
Validation of Artificial Intelligence-Assisted Digital Pathology for the Detection of Breast and Colorectal Lymph Node Metastases: real-life validation to resolve the challenges of translating research into a safe and efficient diagnostic service provision.
IRAS ID
306215
Contact name
Federico Roncaroli
Contact email
Sponsor organisation
University of Manchester
Duration of Study in the UK
1 years, 6 months, 0 days
Research summary
Cellular Pathology is pivotal to cancer pathways, but services are under severe pressure with a year-on-year increase in complexity and volume of patient cases alongside a continuously reducing workforce. These issues require urgent action to ensure resources can respond to the increase in demand.
Breast or bowel carcinoma typically spread to distant locations including lymphatic nodes. The spread of cancer cells to lymph-nodes associates with poorer patients’ survival and affects patients’ treatment. Lymph-node involvement is assessed by radiologists before surgery, and pathologists under the microscope on tissue removed during surgery. Manual assessment is time-consuming and small metastases can be missed due to tiredness, high workload, and pressure to comply with cancer waiting lists.
Whilst there is evidence in the research setting of the accuracy of software in detecting metastases, little is available of the clinical impact of artificial intelligence (AI) in improving performance of individual pathologists and in saving time and resources.
The project is designed to validate the market ready Visiopharm 90159 Metastasis Detection Application VS90159MDAPP on “real-life” cases using FFPE lymph-node sections from patients operated as standard of care during the project timespan. The pathology slides will be scanned, pseudo-anonymised and annotated by VS90159MDAPP. Any disagreement between pathologists and the application will be investigated. The confidence of pathology consultants in using VS90159MDAPP, the impact of VS90159MDAPP on workflows and diagnostic efficiency will be examined through use of a pseudo-anonymised structured questionnaire. The ATracker Pro app will be implemented to measure the time taken to report cases with and without use of the AI-derived data. Cost of consultant and technical time allocations will be assessed to inform the cost-effectiveness of AI-based algorithms in diagnostic provision.
REC name
East Midlands - Nottingham 2 Research Ethics Committee
REC reference
24/EM/0169
Date of REC Opinion
24 Jul 2024
REC opinion
Favourable Opinion