Segmentation of Sinus Images for Grading of Severity of Sinusitis

Sinusitis is commonly diagnosed with techniques such as endoscopy, ultrasound, X-ray, Computed Tomography (CT) scan and Magnetic Resonance Imaging (MRI). Out of these techniques, imaging techniques are less invasive while being able to show blockage of sinus cavities. This project attempts to develop a computerize system by developing algorithm for the segmentation of sinus images for the detection of sinusitis. The sinus images were firstly undergo noise removal process by median filtering followed by Contrast Limited Adapted Histogram Equalisation (CLAHE) for image enhancement. Multilevel thresholding algorithm were then applied to segment the enhanced images into meaningful regions for the detection and diagnosis of severity of sinusitis. The multilevel thresholding algorithms based on Otsu method were able to extract three distinct and important features namely bone region, hollow and mucous areas from the images. Simulations were performed on images of healthy sinuses and sinuses with sinusitis. The developed algorithms are found to be able to differentiate and evaluate healthy sinuses and sinuses with sinusitis effectively.