AUTOMATED DEEP LEARNING-BASED NETWORK FOR DETECTING COVID-19 FROM A LUNG CT SCAN
Journal of Pharmaceutical Negative Results
; 14(2):1850-1862, 2023.
Article
in English
| EMBASE | ID: covidwho-2241743
ABSTRACT
The COVID-19 disease is a threat to public health around theworld. Early diagnosis and detection will be critical factors in preventing the spread of COVID-19. Computed tomography has a significant role in COVID-19 detection because it gives both fast and best results. Hence it is very significant to develop an accurate and rapid computer-assisted tool for helping clinical experts to identify COVID-19 patients from CT scan images. The project's main objective is to develop an artificial intelligence-assisted tool for predicting the severity of COVID-19 with the help of CT scan images. We introduce a new dataset that contains 47,144 CT scan images from 292 normal persons and 14,346 images from 92 patients with COVID-19 infections. In the first stage, the system runs our proposed image processing algorithm that analyses the view of the lung to discard those CT images inside the lung that are not properly visible. This action helps to reduce the processing time and false detection. Then those chosen images from the CT selection algorithm will be fed to the ResNet50V2 model, so the model becomes able to investigate different resolutions of the image and does not lose the data of small objects. Apart from 152 patients,47 patients have been detected with COVID-19, and 105 patients have been detected as Normal. It shows that the model obtained 97.89% correctness overall and 95.45% along with class with COVID-2019 sensitivity.
Full text:
Available
Collection:
Databases of international organizations
Database:
EMBASE
Language:
English
Journal:
Journal of Pharmaceutical Negative Results
Year:
2023
Document Type:
Article
Similar
MEDLINE
...
LILACS
LIS