Your browser doesn't support javascript.
Towards Framework for Edge Computing Assisted COVID-19 Detection using CT-scan Images
IEEE International Conference on Communications (ICC) ; 2021.
Article in English | Web of Science | ID: covidwho-1559839
ABSTRACT
The ongoing pandemic of COVID-19 has shown the limitations of our current medical institutions. There is a need for research in automated diagnosis for speeding up the process while maintaining accuracy and reducing computational requirements. In this work, an IoT and edge computing based framework is proposed to automatically diagnose COVID-19 from CT scans of the patients using Deep Learning techniques. The proposed method requires less computational power and uses ensemble learning to increase the models' overall predictive performance. In the simulation, it was found that each model performs better in some areas than the other. The proposed scheme uses ensemble learning to take advantage of such an occurrence and achieved an accuracy of 86.2% and an AUC score of 89.8% on the COVID-CT-Dataset. This accuracy is achieved keeping the hardware accessibility in mind by training the models using a labeled dataset of CT-scans of the patients. Unlike other works, we were able to train models on a single enterprise-level GPU. It can easily be provided on the edge of the network, which reduces communication overhead and latency. This work aims to demonstrate a less hardware-intensive approach for COVID-19 detection with excellent performance combined with medical equipment and help ease the examination procedure.
Keywords

Full text: Available Collection: Databases of international organizations Database: Web of Science Type of study: Prognostic study Language: English Journal: IEEE International Conference on Communications (ICC) Year: 2021 Document Type: Article

Similar

MEDLINE

...
LILACS

LIS


Full text: Available Collection: Databases of international organizations Database: Web of Science Type of study: Prognostic study Language: English Journal: IEEE International Conference on Communications (ICC) Year: 2021 Document Type: Article