Research Of OCT Image Segmentation Algorithm Based On Region Growing Method

Medical image segmentation is a complex and key step in the field of medical imageprocessing and analysis, can provide reliable basis for clinical diagnosis and pathologystudy. Because of the complexity of the medical image itself, the general imagesegmentation method is difficult to directly applied in medical image segmentation. Atfirst, this paper studied the medical image segmentation algorithm, including:pretreatment of images before segmentation, classical segmentation algorithms and theiradvantages and disadvantages, application of images after segmented and evaluation ofsegmentation algorithms, and made a more comprehensive introduction of medical imagesegmentation algorithms. Then analyzes the principle, advantages and disadvantages aswell as application in medical image segmentation of the classic segmentation algorithm"region growing method". Finally through the analysis of the nature of medical OCTimage, this paper puts forward a improved algorithm based on region growing, thisalgorithm can realize rapid and automatic segment such medical OCT images. The maincontent of this topic research and the innovation points are as follows:(1) Compared with other segmentation algorithms, chose region growing algorithmto deal with this kind of OCT images. Through the analysis of such OCTimage，found that the target mucous membrane layer have no obvious boundarybetween the other mucous membrane levels, only some difference on the graylevel gradient. So this paper chose the threshold value method to segment theOCT images. The grey value and difference of the target area is not so big, sothis paper chose region growing method to segment such medical OCT images.(2) Improved region growing algorithm, which can automatically segment the OCTimages. Through the analysis found that the grey value of the mucousmembrane layer in the target area is higher than other parts. If we use regiongrowing method to extract the whole of mucous membrane layer then thestarting point must be choice on the grey value deeper layer of mucousmembrane.this study improved the way of the choice of starting point, so thealgorithm could automatically select growth starting point and realize theautomation of such OCT image segmentation. (3) Improved growth criterion of the region growing algorithm, and improved thesegmentation accuracy. Based on the characteristics of this kind of OCT image,this research improved the growth rules of region growing, significantlyimproved the precision of image segmentation, and over segmentation and owesegmentation phenomenon ware greatly reduced.(4) Using linear compression method to compress image, the image data reduction,and improve the processing efficiency. One of the feature of is OCT highresolution, but bring very great difficulty to the image post-processing, weadopts the image compression method to simplify the redundant information,reduction after image processing, greatly reduces the running time of thealgorithm, improved the efficiency.