Skip to main content

Detecting plaque in carotid arteries from ultrasound images

Research Team: Dr. Jacquita Affandi, Prof. Christopher Reid
CIC Specialist: Dr. Kevin Chai

Cardiovascular disease remains as the number one cause of deaths in the world. In Australia alone, one in four of deaths is caused by cardiovascular disease with 1,600 hospitalisations per day, and more than $5 billion spent on providing health care services to patients. Atherosclerosis is the main underlying cause of cardiovascular disease. This is marked by plaque deposits that are made up of fat, cholesterol, calcium and other substances built up on the walls of carotid arteries. Over time, plaques harden and narrows the opening of the arteries, thus restricting blood flow. These plaque deposits may break open and form a thrombus or a blood clot, which can then lead to a stroke event.

Ultrasound is a technique that can help locate and identify plaques in arteries. However, there are challenges when using ultrasound in practice. Firstly, it is time intensive for clinicians to manually review ultrasound images. This not only limits the number of patients that can be serviced by clinicians but can also result in missed detections due to human errors. Secondly, the identification of plaques can be subjective and there is an opportunity to develop more objective method for identifying plaque candidates. Lastly, early detection of plaques allows for more time to be spent by clinicians on prevention and treatment leading to better patient health outcomes.

To address these challenges, Dr. Jacquita Affandi and Prof. Christopher Reid from the Centre for Clinical Research and Education (CCRE) collaborated with Dr. Kevin Chai from the CIC to develop a machine learning algorithm to automate the plaque detection process. A preliminary experiment was conducted and the algorithm achieved 100% mean Average Precision on ultrasound images/videos from 61 patients in the PACIFIC study led by the CCRE. The project has received interest and was shortlisted in to the second stage of 2020 Curtinnovation awards and received a scholarship for the Curtin Ignition program to assist in translating the developed algorithm into healthcare practice.