Ultrasound AI of thyroid nodules to enhance preoperative diagnosis
Research type
Research Study
Full title
Improving preoperative diagnosis of thyroid nodules by developing ultrasound artificial intelligence (AI) decision support system
IRAS ID
284038
Contact name
Sidhartha Nagala
Contact email
Sponsor organisation
Royal Berkshire NHS Foundation Trust
Duration of Study in the UK
1 years, 0 months, 1 days
Research summary
Thyroid cancer is the most common malignant endocrine tumour, with an annual incidence in the UK of 5.8 per 100,000 per year. Thyroid nodules may have benign or malignant pathology. Current diagnostic work-up involves ultra-sound plus or minus needle biopsy. However despite repeated needle biopsies, up to 30% of lumps yield indeterminate test results. The risk of malignancy within these indeterminate lumps is 20%–30%. Patients are often recommended diagnostic surgery to rule out cancer. Better preoperative diagnosis would reduce unnecessary operations and improve management of patients with thyroid lumps. More research is needed to explore new techniques that discriminate between malignant and benign thyroid lumps.
This project aims to build upon the chief investigator’s University of Cambridge PhD thesis which showed promising novel results using texture analysis and computer-aided diagnosis for thyroid nodules.
Our vision is to create commercially available artificial intelligence (AI) decision support system that predicts the histology of thyroid nodules using non-invasive ultrasound.
This pilot project will aim to use AI and machine learning computer techniques to automatically classify thyroid nodules to the current ultrasound classifications (ACR-TIRADS) and and postoperative histology.
REC name
North West - Haydock Research Ethics Committee
REC reference
21/NW/0167
Date of REC Opinion
19 May 2021
REC opinion
Favourable Opinion