OPG Images Automatic Segmentation and Feature Extraction for Dental Lesion Diagnosis Purposes

Dental lesions which affecting jawbone can be detected using Orthopantomogram (OPG). Cysts lesions and tumors of OPG region are a major subject for health care concerned. Detection of these lesions is important for accurate diagnosis of these lesions in early stages that has a positive consequence on prediction of the disease. Cystic lesions if not detected and treated early may possibly lead to tumors. Generally, image analysis and techniques are required in order to improve the OPG image quality for the purpose of obtaining a precise dental lesion diagnosis using an efficient hybrid filtration of noise reduction via adaptive median filter (AMF) and discrete wavelet transform (DWT) and contrast enhancement process via mathematic morphology and contrast limited adaptive histogram equalization (CLAHE). Then, automatic segmentation of sequential processes for detection object with aid using Otsus thresholding and canny filter. Thus, texture features extraction task will be accomplished from these segmented objects which are First-order statistics texture (FO), Gray level co-occurrence matrix GLCM and Gray Level Run Length Matrix (GLRLM). Such feature will lead to a better organization result for a precise diagnostic. This paper presents an algorithm for automatic segmentation of OPG images to detect suspicion abnormal segmented regions of lower jaw using OPG. The proposed algorithm consists of sequences steps which are image enhancement, segmentation and feature extraction.