An Efficient Fast Marching Algorithm For Dental Cyst Segmentation In Digital Radiographs

Dental radiographs have been widely used by dentists in monitoring the progress of the periodontal defect treatment. And more importantly, digital dental X-rays make computer aided dental X-rays analysis possible and convenience. In digital dental X-rays the regions of the cysts are of low contrast and the pixel intensity distribution is not homogenous so cyst segmentation is a complex problem in digital dental radiography. To efficiently analyze the cyst images we need an automatic segmentation, which can simplify the data analysis and could be used for further feature extraction and computerized diagnostic support system design, in particular to detect potentially precancerous lesions. This paper introduces a semi-automatic segmentation and Interface propagation based method called Fast Marching Method (FMM), applied to find the boundaries of the suspicious oral lesions and for delineation of various types of dental cysts. The fast marching method is associated with the boundary value problem, where the propagation strictly expands or contracts. The fast marching is robust and efficient method uses this interface concept for solving the complex problem of cyst segmentation and helps the dentists in diagnosis. But FMM requires only the seed point to start with and need less human intervention and supervision. So this fast marching method is an efficient, robust in segmenting the Dental Cyst. The results are compared with other segmentation techniques like watersheds, Thresholding, Live-wire, Template matching, Level set segmentation. Experimental results show that the proposed method is competitive or better than the other traditional methods for cyst segmentation.