Abstract

More and more, CT imaging is needed in the pre-diagnosis of COVID-19. The accumulation of data gives it more precision. However, in differential diagnosis, it is necessary to take into account peripheral pathologies such as pulmonary edema, infectious bronchiolitis or bacterial pneumonia. In the diagnosis susceptible to COVID-19, it is essential to identify its specificities in terms of location and characteristics. Signs of this show up on imaging as a moderate increase in lung parenchyma density secondary to edema. It is often bilateral and multifocal, is located on the periphery, inferior and posterior. Rarely, these signs are atypical. In order to properly interpret these images, it is necessary to take into account all thoracic pathologies for the elimination of differential diagnoses. It seems obvious that the system is much more complex to arrive at an exact reading and to pronounce. This study deals with the analysis of the specificities of CT images in relation to the patient’s status in terms of thoracic co-morbidity. As the system is complex, analysis using artificial intelligence techniques is proposed. The principles of fuzzy inference are applied. Data relating to the patient’s history and the nature of the recorded image are considered as input variables to the system. The degree to which COVID-19 has been confirmed or denied is the output variable of the system. The fuzzy analysis makes it possible to compensate for the uncertainties associated with the process and therefore the diagnosis will be as precise as possible. It can be seen as a diagnostic aid.

Keywords

Tomography CT, Covid-19, Data analysis, Fuzzy logic