SEGMENTATION APPROACH AND ANALYSIS OF THERMAL IMAGES FOR DIAGNOSING SINUSITIS

This paper aims to detect the presence and analysing the type of paranasal sinusitis by using thermal imaging. This is done by analysing the four different sinusitis tissues in the face: the maxillary sinus tissues, the ethmoidal sinus tissues, the sphenoidal sinus tissues and the frontal sinus tissues. The optimal selection of the segmentation is important in the development process of a preliminary screening technique to detect the presence of accumulated sinus fluid in these four regions, which give rise to a significant temperature difference. This temperature difference is a sign of anomaly, indicating the presence of paranasal sinusitis. The thermal camera used is the FLIR One Pro model, and the thermal image taken is then subjected to segmentation techniques in MATLAB software. The segmentation technique used in this paper is multilevel threshold segmentation. This analysis provides insight on the relationship of sinusitis location in the face along with its type and severity. These thermal images can be used in regular check-ups to diagnose the variations in the subject’s condition over a period of consistent medication suitable for PNS and it can be stored in a report format in a database for easy access of both- the subject and the medical practitioner. The analysis of results are carried out by a GUI developedin which the thermal image is given as input, and the proposed GUI automatically detects the region of interest and gives the diseased region as output.