Liver cancer detection using recent advances in deep learning

Approximately half of todays population will be affected by cancer during its lifetime. Knowing that the earlier cancer is detected, the better the life expectancy, accurate radiological detection can make the difference between life and death. Primary liver tumor, known as hepatocellular carcinoma (HCC), and secondary liver tumor (metastasis) represent, in aggregate, the second most common cause of cancer-related mortality in North America. Existing Computed Assisted Diagnosis (CADx) software use traditional image analysis techniques and as such their applications are mainly oriented towards type of cancer that show specific imaging characteristics (e.g. lung nodules, colon polyps and breast micro calcifications). The goal of this project is to innovate in the field of automatic medical image analysis by using the latest developments in artificial intelligence, computer vision and machine learning. Using readily-available data of sufficient quantity from the liver through extensive datasets hand-segmented, we plan to develop a framework that will detect and segment tumors in a liver segmented from CT and MRI images. This approach opens new prospects for medical imaging applications, by proposing of models that exhibit high accuracy with minimal human intervention, broadening their use and ultimately leading to improved patient care.

Faculty Supervisor:

Christopher Pal

Student:

Mohammad Havaei

Partner:

Global Imaging On Line

Discipline:

Computer science

Sector:

Information and communications technologies

University:

Program: