International Journal on Advanced Science, Engineering and Information Technology, Vol. 11 (2021) No. 2, pages: 595-602, DOI:10.18517/ijaseit.11.2.14143

Automated Identification Model of Ground-Glass Opacity in CT-Scan Image by COVID-19

Jufriadif Na'am, Feriantano Sundang Pranata, Rahmat Hidayat, Arrianda Mardhika Adif, - Ellyzarti


Ground-Glass Opacity (GGO) is an object found in the thorax due to infection. This object interferes with the normal function of the thorax in breathing. The characteristic of GGO has slightly lighter turbidity compared to normal thorax tissue on radiological images, so it is very difficult to identify it precisely. This study aims to identify the GGO pattern and find the exact area of the CT-scan image of COVID-19 sufferers. The data tested were 34 images from 34 different patients. The image was taken using CT-Scan equipment with the tube model 46274891G1 axially. Each patient is taken one image with the reading position right above the chest using the file format Joint Photographic Experts Group (jpg). An automatic image processing model developed in this study uses several interrelated and continuous technical steps; Image Enhancement, Convert to Binary Image, Morphology Operation, Image Inverted, Active Contour Model, Image Addition, Convert Matrix to Grayscale, Image Filtering, Convert to Binary Image, Image Subtraction and Region Properties. The results of this study can identify GGO in all patient test images, where each patient has GGO. The smallest area of GGO was 3.9%, and the highest was 34.2% of the total thorax area. This level of comparison is greatly influenced by the severity of the COVID-19 virus patient. This area of GGO weakens the normal function of the thorax in the respiratory process of the patient. Thus, this research can be used as a model recommendation in identifying thorax damage due to COVID-19 very well in following up on more intensive treatment in the future.


COVID-19; Ground-Glass Opacity (GGO); CT-scan; thorax; automated identification.

Viewed: 319 times (since Sept 4, 2017)

cite this paper     download