Classifying the lung images for people infected with COVID-19 based on the extracted feature interval

ABSTRACT

This study proposes a new method to classify image data for two groups and effectively apply it in identifying people infected with COVID-19 based on their lung image. First, we extract each image into a two-dimensional interval based on the Gray Level Co-occurrence Matrix (GLCM). Next, find the prior probability based on the fuzzy relationship between the classified element and the established groups. Finally, combining the above improvements, we propose a new principle similar to the Bayesian method for classification. An image is classified into a group if it has the most significant value for prior probability and the smallest value for the overlap distance between the representative interval for the image and the groups. Applying the set of lung X-ray images to distinguish people infected with COVID-19, the proposed algorithm has given outstanding results compared with many well-known methods. The result also shows that this research can be applied in practice, and has the potential for different fields.

KEYWORDS:

• Classification
• image set
• overlap distance
• prior probability

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research is funded by Vietnam National University Ho Chi Minh City (VNU-HCM) under grant number C2022-26-07.