Development of AI-enabled Tools for Advanced Clinical PET/CT Imaging of Cancer Patients

Over 200,000 new cancer cases are diagnosed in Canada each year. With imaging using an appropriate modality, many types of cancer that manifest as solid tumors can be detected, treated or managed effectively. Positron emission tomography (PET) combined with computed tomography (CT) is the primary imaging modality in a range of cancer types. Scientific studies have determined that measuring the size, shape, and texture of tumors from PET/CT images can help identify patients at high risk of early cancer recurrence, or for whom the standard treatment may fail. Nevertheless, the process of image reading in the clinic remains largely qualitative, since manual tumor delineation by radiologists can significantly reduce patient throughput and increase scan wait times. The goal of this project is to design artificial intelligence (AI) tools to assist radiologists and scientists in automatic detection and delineation of tumors in PET/CT images. In collaboration with Microsoft, we will deploy such tools in the cloud, and make them available to practicing physicians and cancer researchers at BC Cancer. The expected benefits of AI-enabled PET/CT image analysis include faster diagnosis, more personalized treatment plans, improved treatment outcomes, and reduction of healthcare costs in Canada.

Faculty Supervisor:

Arman Rahmim

Student:

Ivan Klyuzhin

Partner:

Microsoft Canada

Discipline:

Other

Sector:

University:

University of British Columbia

Program: