Using AI to improve PET/CT image interpretation in Lymphoma management
Positron Emission Tomography with Computed Tomography (PET/CT) can be used for cancer patients to determine the extend of the disease at staging, to monitor treatment at interim time points or to assess remission after treatment has been completed.
Charity partner: NCRI
PET scanners work by detecting the radiation given off by an injected radio tracer which collects differentially in different parts of the body. The resultant images are essential for accurate diagnosis, staging and treatment planning of many cancer types.
State-of-the-art clinical PET/CT reading software enables the visualisation and quantification of the images, providing information which is input in the clinical diagnostic process.
Hodgkin and non-Hodgkin lymphoma scans can prove difficult to report, as they present highly complex PET uptake patterns ranging from single site compact lesions to multiple lesions both above and below the diaphragm and additional bone marrow and spleen involvement. Depending on the difficulty and the stage of the case, it can take between 15 and 60 minutes to process a lymphoma PET/CT case, with reports varying substantially depending on the experience of the reporter.
AI has the potential to make a significant difference in healthcare. Mirada is a global leader in this area, having developed DLC Expert, the first FDA cleared deep-learning product for contouring organs at risk in radiotherapy.
Building on its established partners with the Alliance Medical Group and the NHS PET scanning and reporting programme, Mirada Medical will investigate the role that AI and deep learning can play in diagnostic nuclear medicine.
Mirada XD, an integrated software application which allows efficient reading of PET/CT, is already used to read more than half of the PET/CT scans in England and Wales. This project will aim to integrate AI into Mirada XD, thereby improving the efficacy of cancer diagnosis and treatment response assessment.