User Tools

Site Tools


projects:year10:10a.002.tau_wp1

10a.002.TAU_WP1 - COVID-19 Severity Grading Using Chest X-Ray Images

Project - Summary

Chest X-ray (CXR) imaging has been used to automatically diagnose the coronavirus disease 2019 (COVID-19). Despite the risk of overfitting, most of the previous studies employed Deep Learning models on sparse data. Additionally, prior research has shown that deep networks are unreliable for classification since their conclusions could come from unrelated regions on the CXRs. In order to accomplish detection by segmenting COVID-19 pneumonia for a trustworthy diagnosis, in this project, we propose Operational Segmentation Network (OSegNet). This project expands the largest COVID-19 CXR dataset: QaTa-COV19 with 121,378 CXRs including 9258 COVID-19 samples with their corresponding ground-truth segmentation masks that are publicly shared with the research community. Accordingly, QaTa-COV19 dataset addresses the data scarcity encountered in training and particularly in evaluation.

Project - Team

Team Member Role Email Phone Number Academic Site/IAB
Moncef Gabbouj PI moncef.gabbouj@tuni.fi 358 40 073 6613 Tampere University
Serkan Kiranyaz Co-PI serkan.kiranyaz@tuni.fi 97 43 063 5600 Tampere University
Aysen Degerli Researcher aysen.degerli@tuni.fi 358 46 521 9737 Tampere University
Ozer Devecioglu Researcher ozer.devecioglu@tuni.fi N/A Tampere University
Christian Sundell Project Mentor Christian.Sundell@tietoevry.com NA TietoEVRY
Iftikhar Ahmad Project Mentor iftikhar.ahmad@tietoevry.com N/A TietoEVRY

Project - Novelty of Approach

See “Project Summary” section above.

Project - Deliverables

Deliverables
1 Integrating OSegNet model into the http://qatacov.live/ website

Project - Benefits to IAB

The proposed system will provide a time-efficient and robust diagnosis, which will be a life-saving tool in the pandemic. The system can easily be set up in health care centers, airports, and hospitals. The largest CXR dataset for the severity grading of COVID-19 will be publicly shared with the research community. Thus, the dataset will be the benchmark in this domain and used in the upcoming research studies in this domain. Next, Self-ONN that is implemented for the image segmentation task can be used in any other application containing a segmentation task.

Project - Documents

projects/year10/10a.002.tau_wp1.txt · Last modified: 2022/10/03 14:51 by sally.johnson