Safe People
Lead applicant organisation name	University of Birmingham

Safe Projects
Project title	Three-dimensional reconstruction of anal fistula images using machine learning
Lay summary	An anal fistula is an abnormal communication between the anal canal and the surrounding skin, resulting in pain, swelling and discharge of pus. It affects 1 in 5,000 people and can be a chronic disease. Surgery aims to cure but sometimes is unsuccessful. There is a risk of damaging the delicate sphincter muscles, causing incontinence. Most often, anal fistulas are single tracts but they can be complex with multiple tracts and branches. If important branches are neglected when performing surgery, the treatment will fail. A number of different operations exist to treat anal fistula; Selecting the correct operation depends on the anatomy of the fistula and its complexity.
Public benefit statement	Automatically generating three-dimensional images for anal fistula patients significantly improves the view of the abnormal tunnels (fistula tracts) in relation to crucial structures, such as the sphincter muscles (which control the opening and closing of the anus) and the pelvic floor (muscles that support the pelvic organs). An anal fistula is an unintended passage that forms between areas in the anal region. This advanced imaging supports surgeons with clear planning and helps explain to the patient their condition and proposed treatment. It represents the first tangible benefit for patients at University Hospitals Birmingham NHS Foundation Trust. Future plans include expanding the project to incorporate MRI scans from various units and different scanner models to ensure the AI model’s consistency. There is also potential to develop an AI algorithm that predicts surgical success and the risk of post-operative incontinence based on imaging analysis.
Latest approval date	15/01/2024

Safe Data
Dataset(s) name	PATHWAY Research Data Hub: PWY001 dataset

Safe Setting
Access type	Data Licence Agreement between the collaborating organisations using secure and controlled data analysis environments.

Safe Outputs
Link	Not yet Published