Methods of processing medical images for the detection of pathologies in craniofacial surgery

Abstract

Analyzing the algorithms for filtering digital images, we have pointed out dimensional image processing techniques. We have studied the performance of smoothing filters, the work of differential operators, and the efficiency of the Canny method. We have determined the peculiarities of applying the algorithms for identifying CT scans boundaries. We have researched the possibility of applying the Serra mathematical morphology operations for determining the most significant and muting uninformative image objects. While specifying the boundaries, we have visually assessed the quality of noise reduction operations, the efficiency of various operators and techniques as well as the rate of performance and the minimum average quadratic deviation. As a result of the study, we have suggested an algorithm for obtaining the morphological features of pathologies at the stage of diagnostics in craniofacial surgery. The essence of the algorithm lies in distinguishing the original information from the resulting model. Obtaining the original information means getting rid of the noise from a set of tomographic slices with the Gaussian filter, identifying the most informative image objects by means of the operation of mathematical morphology Close, and, depending on to the type of pathology, outlining the shot boundaries with the Sobel, Roberts and Laplace differential operators as well as by the Canny method. The model construction includes mirroring the healthy side and processing the obtained information with the help of the above listed methods.