Research And Application Of Foveal Avascular Zone Detection Technology Based On OCTA Image

With the progress of science,modern medical imaging technology plays a more and more important role in medical diagnosis.In recent years,a new optical imaging technology after X-CT and MRI technology has emerged,optical coherence tomography(OCT),it can realize non-invasive tomography of biological tissue with high resolution and has wide application prospect.At present,OCT technology has been widely used in the treatment of many diseases,especially in the treatment of ocular diseases.OCT Angiography(OCTA)is a fast,non-invasive,novel blood flow imaging technology.It has the advantages of high resolution,fast scanning speed and quantifiable blood flow.It has important research value and the prospect of clinical application.This paper presents an algorithm for detecting the Foveal Avascular Zone(FAZ)in the OCTA image,which is based on the active contour model.This study realized the automatic detection of macular foveal vascular zone for eye images,and the valuation of FAZ metrics for diagnosis of diabetic retinopathy.The main research work is as follows: 1.Research on the basic theory of active contour model.The regional characteristics of the FAZ in the OCTA image are carefully studied,including the gray and geometric characteristics.Reading lots of relevant literature of image segmentation methods and classifying them.According to the regional characteristics of FAZ and the analysis of the characteristics of active contour models,generalized gradient vector flow(GGVF)is applied to detect FAZ region.2.The realization of the fully automatic FAZ detection algorithm.The detection process includes 3 main parts:(1)preprocessing to reduce residual motion artifacts and background noise and acquire a binary angiography.(2)get the initial GGVF search boundary;(3)applying the GGVF active contour model.3.Statistical analysis of FAZ Metrics.FAZ metrics include two conventional FAZ metrics(FAZ area coefficient(FAZ area),FAZ roundness coefficient(AI)),and two novel FAZ metrics(sector irregular coefficient(STD4),ring irregularity coefficient(NR300)).Among them,the novel FAZ metrics have took the effect of normal variation in FAZ size into account partially.The statistical analysis,which includes the variance analysis of the FAZ metrics between DR groups,the correlation between the FAZ metrics and DR groups,the sensitivity of FAZ metrics to distinguish the DR groups.4.The conclusion of the experiment.The GGVF algorithm tested in this study can accurately and reliably detect the FAZ.However,although FAZ metrics can provide clinically useful information regarding macular ischemia,and possibly visual acuity potential,EAA measurements may be a better biomarker for DR.The innovations of the paper are as follows:(1)by introducing the GGVF active contour model,the high-precision detection of FAZ is realized.This achievement can assist clinical FAZ determination and shorten processing time.(2)two novel FAZ metrics in diabetic retinopathy were proposed.Statistical analysis shows that FAZ measurements can provide useful information for the research of macular ischemia and visual function.The shortcomings of the paper are as follows:(1)In the FAZ detection algorithm,the detection results of a small proportion test cases(5%)aren't satisfying.The reason is that the GGVF active contour algorithm based on parameter is more dependent on the edge map.Therefore,FAZ boundaries of some participants are too deep or narrow to show good detection results.(2)the number of participants in this study is moderate,and cross-sectional design and research use age and sex mismatched control group.This makes the study have some limitations.All in all,the results of this study can realize the automatic detection and quantitative analysis of FAZ area in most diabetic retinopathy patients.It can not only provide accurate and reliable information for the doctor's related diagnosis,but also can promote the further clinical application and popularization of OCTA related detection technology.