Reza Rawassizadeh

Dr. Reza Rawassizadeh and Team Produce Deep Learning Model to Identify COVID-19 Infection

Reza Rawassizadeh
Associate Professor, Computer Science
PhD, University of Vienna (Austria) Master of Computer Science Management, Vienna University of Technology (Austria) Bachelor of Computer Engineering, Azad University of Tehran, Central Branch (Iran)

What is your area of expertise?
I am active in two areas, ubiquitous computing and machine learning. My contributions to ubiquitous computing focus on building wearables and designing robot algorithms, applications that can monitor users’ health.

My contributions to machine learning are focused on developing standalone, energy-efficient algorithms that operate independently from any network and on-device.

Please tell us about your work. Can you share any current research or recent publications?

One research effort that has been implemented in a very short amount of time is our deep learning model to identify COVID-19 from computed tomography (CT) images, which we call “CovidCTNet.” You can find the e-print on arXiv at https://arxiv.org/abs/2005.03059.

Through collaboration with many colleagues from BU MET’s Health Informatics programs and Health Informatics Research Lab (HILab)—including Dr. Lou Chitkushev and Dr. Guanglan Zhang—as well as with researchers from other local and international institutions, we have created a model that can accurately identify and distinguish COVID-19 infection from other pulmonary diseases. It even out-performs human radiologists in its accuracy. Below, you can see a small picture of our algorithm and how it can extract the infected area from the lung. On the left side is the 3D image of the lung, and on the right side is the 3D image of infection distribution in the lung.


How does the subject you work on apply in practice? What is its application?
One of the emerging fields in computer science is “computer-aided diagnosis” (CAD), which might make some shifts in traditional medical science. For example, telemedicine has received lots of attention recently. In the post-COVID-19 era, physicians might rely more on patients’ digital data collected from sensors, rather than traditional communications. Therefore, ubiquitous devices—such as care robots—that continuously follow the user and track their vital signs, might get more attention.

What courses do you teach at MET?
I am teaching Web Analytics and Mining (MET CS 688) and Advanced Programming Techniques (MET CS 622).

Please highlight a particular project within these courses that most interests your students.
Last semester in the Web Analytics and Mining course we collected data from different media, including Google News, Bing News, PubMed, DBLP, Indiegogo, and Kickstarter. Two of our former students have analyzed how “wearable technologies” have changed and what will be the future direction of these technologies. We have lots of interesting findings, which we have summarized in a scientific paper, which is currently under review.

 

View all posts