Congratulations to Vasileios Magoulianitis for passing his defense today. Vasileios’ thesis is titled “Transparent and Lightweight Medical Image Analysis Techniques: Algorithms and Application.” His Dissertation Committee includes Jay Kuo (Chair), Justin Haldar, and Qifa Zhou (Outside Member). The Committee members were pleased with the breadth and depth of Vasileios’ thesis. The MCL News team invited Vasileios for a short talk on his thesis and PhD experience. Here is the summary. We thank Vasileios for his kind sharing and wish him all the best on his next journey. A high-level abstract of Vasileios’s thesis is given below:
Thesis Title: Transparent and Lightweight Medical Image Analysis Techniques: Algorithms and Applications
The thesis contains two main research topics:
Nuclei segmentation in histopathological images and Prostate Cancer (PCa) from Magnetic Resonance Imaging (MRI). On the one hand, histopathological images are meant to detect and grade cancer. Toward this end, nuclei segmentation is a cornerstone task to reveal the molecular profile of the tissue. Three self-supervised solutions have been introduced: (1) CBM, which uses a parameter-free pipeline using thresholding, (2) HUNIS where a novel adaptive thresholding and false positive reduction module are proposed and (3) Local-to-Global NuSegHop where a novel feature extraction method is proposed. On the other hand, PCa-RadHop pipeline is proposed for prostate cancer detection from MRI, achieving a competitive performance with a model size orders of magnitude smaller than other Deep Learning based models.
PhD Experience Sharing:
The PhD journey within USC and MCL has been an experience I will remember for a life. The first years in the PhD I had to take many courses to build my theoretical insights and achieve the first milestone to pass the screening exam which required a very rigorous preparation. In my entire PhD life, I had two research projects going on alongside. Regarding the MRI prostate cancer detection problem, I was given the chance to collaborate with physicians and understand more about the applications in healthcare and look at the problems I was solving from the end user’s standpoint. The second pillar of my research was the nuclei segmentation project. Throughout these projects I had the opportunity to collaborate with several students, learning a lot from our interaction about research and programming. The different responsibilities in MCL, ranging from mentoring new students up to delivering research tasks for the projects helped me to grow my independent thinking and take on initiatives. Moreover, since a PhD student needs to switch among different tasks every week, one of the most valuable experiences is the time management and how to focus on research problems among other tasks.
In MCL the team spirit sounds loud, and students are very supportive to each other. I have found particularly useful to be able to use a new technology developed in MCL in favor to my research. Professor Kuo is a very supportive advisor and the best ally to the PhD battlefield. I am very grateful to him and all the MCL members who have helped over the course of my PhD. All in all, the PhD experience in USC was a magnificent journey and despite its difficulties, the feeling of accomplishment and knowledge growth is definitely rewarding and the MCL experience has lived up to my expectations.
