A Transfer Learning Model for COVID-19 Detection with Computed Tomography and Sonogram Images
6th International Conference on Wireless Communications, Signal Processing and Networking, WiSPNET 2021
; : 80-83, 2021.
Article
in English
| Scopus | ID: covidwho-1255053
ABSTRACT
The community spread of COVID-19 has almost touched the corners of the world. So it is very essential to step towards the much earlier diagnosis of COVID-19 infection. In general, early stage diagnosis techniques mainly consider Chest X-Rays, Computed Tomography and Ultrasound videos as well as frames. Visually identifying and examining these clinical images in case of any hidden abnormalities seems to be a challenging and also time consuming task, wherein a huge amount of information needs to be processed in a limited time period given. The concept of Transfer Learning on medical imaging is currently emerging and shows a lot of research potential in terms of dealing with scarce medical images. Here, we proposed a Transfer Learning model and keras model for CT images and a separate transfer learning model for ultrasound images. As a result, the CT Transfer learning framework outperforms the CT based keras Baseline model with an ascend accuracy from 76% to 82.2%. Furthermore, our Ultrasound based Transfer learning model almost tied with an accuracy of 89% as a good performing solution. © 2021 IEEE.
Full text:
Available
Collection:
Databases of international organizations
Database:
Scopus
Language:
English
Journal:
6th International Conference on Wireless Communications, Signal Processing and Networking, WiSPNET 2021
Year:
2021
Document Type:
Article
Similar
MEDLINE
...
LILACS
LIS