[View Image]

Charles Dietzel

DENSE MRI Scan Auto-Segmentation Project

Faculty mentor: Dr. LaMont Cannon

Department: Center for the Study of Biological Complexity

In a nutshell: Currently, certain kinds of MRI scans of the Aorta involve large amounts of complex computer analysis [and] certain steps needed to prepare each MRI scan for computer processing cannot be done automatically, requiring many hours of manual labor. I am developing a computer program to automate this process and enable more accurate MRI scans of the Aorta.

In a bigger shell: Aortic Aneurysms are a medical issue...currently diagnosed by using an MRI scan to measure the diameter of the Aorta, a flawed and somewhat inaccurate method...Therefore, a better approach for diagnosing the true risks of Aortic rupture in each patient would be to measure the physical strain placed on the Aorta as the heart beats. An MRI technique known as Cine DENSE analysis can calculate this Aorta strain, but it requires each frame of the cardiac MRI scan to be segmented to isolate only the pixels containing the Aorta wall. This segmentation process is both error-prone and very time consuming, [preventing] this diagnostic method from being used in the medical industry. Therefore, the development of an algorithm capable of performing this segmentation process accurately and automatically enables much greater accuracy when diagnosing the risk of Aortic aneurysm rupture, [allowing] for more appropriate treatments to be applied more often, improving patient outcomes.

End of year goal: Develop an automated way of identifying the pixel boundaries of the Aorta from an MRI scan [enabling] more accurate diagnosis of the risk of Aortic Aneurysm rupture

A tip for others: Learn about image processing, linear algebra, and MATLAB programming

View graphic versionView graphic version