Artificial Intelligence to grade oral epithelial dysplasia
Research type
Research Study
Full title
Artificial Intelligence to grade oral epithelial dysplasia and predict malignant transformation
IRAS ID
251234
Contact name
Syed Ali Khurram
Contact email
Sponsor organisation
Sheffield Teaching Hospitals NHS Foundation Trust
Duration of Study in the UK
2 years, 0 months, 1 days
Research summary
Oral epithelial dysplasia (OED) is a term used to describe precancerous changes in the lining of the mouth. Clinically it may present as a white patch (leukoplakia), a red patch (erythroplakia) or a mixed white-red patch (erythropleukoplakia). Tissue from the mouth needs to be seen under the microscope by a pathologist to establish the presence and grade of pre-cancer (or dysplasia) which informs future treatment decisions.
OED are scored (or graded) into ‘low’, ‘moderate’ and ‘high’ categories to predict the risk of oral cancer development. Oral cancer is one of the top ten most common cancers in the world with increasing incidence and worsening survival, therefore early detection and an accurate diagnosis is essential to guide treatment ensuring the best outcome.
Conventional grading of OED is subjective taking into account a wide range of subjective features open to individual interpretation (such as cell shape, size, abnormal dividing cells, architecture etc.) meaning that it is difficult to get a consensus between different pathologists. Since the decision to watch or surgically remove OED is dependent upon the assigned grade, this variability between pathologists can have significant clinical implications with regards to patient treatment, cancer development and follow up.
It has now been shown that after supervised training (Machine learning/ML), computers can be trained to identify patterns in tissues themselves (Artificial Intelligence/AI) to help diagnosis and patient care. This method has been shown to be effective in detecting key features consistently from cases considered difficult and challenging even by experienced pathologists. Research shows that ML/AI can remove subjectivity and variability in cancer tissue analysis ensuring standardisation and a numerical output that can be vital in informing treatment decisions.
This study will aim to determine whether AI can improve the detection and grading of OED (oral pre-cancer) and predict potential of cancer development.
REC name
West Midlands - Edgbaston Research Ethics Committee
REC reference
18/WM/0335
Date of REC Opinion
19 Oct 2018
REC opinion
Favourable Opinion