الفهرس 
Acknowledgements
locta techniquesOnly 14 pages are availabe for public view
 |  | from | 125 |  |  |
| | | from | 125 | | |
Abstract
6.1 Original aimsThis thesis aimed to (1) evaluation of image quality and validation of the automated image quality indices using simple manual grading system, (2) evaluation of new FAZ parameters and distribution of capillary dropout in patients with retinal vascular diseases using the new technology of projection artifacts removal. 6.2 Summary of key findings and implications In this thesis, we demonstrated a simple, easy to learn manual grading system of the superficial capillary plexus OCTA images. Using this system as a quality measure, we showed that SQI is superior to SSI in discriminating between acceptable and unacceptable images. Our results provide some evidence for using a higher index than manufacturer’s recommended minimum threshold for SQI (≥ 7 as compared to ≥ 6) in the current Phase 7.0 software might be preferable. The high rates of unacceptable OCTA images in patients with retinal vascular diseases may limit the utility of OCTA-derived vascular metrics as a clinical trials endpoint measure. Following from the conclusion of our initial results and using the new Angioanalytic software we examined OCTA-derived FD-300 , FAZ parameters and distribution of capillary dropout in some retinal vascular diseases. We found that OCTA-derived FD-300 and FAZ parameters may be useful in predicting visual potential. Each type of retinal vascular diseases shows a unique profile of SVC and DVC involvement. FAZ perimeter, AI and FD-300 may be used as a marker of DR severity even at stage R1. Further structure-function correlation studies are needed to examine the physiological basis of the associations observed in this study. 6.3 Appraisal of limitations and recommendationsThere are some limitations of the study. First, a portion of the images graded were taken from a retrospective cohort of subjects who had routine clinical imaging. We only had selected retinal vascular diseases and the imaging procedure was done by routine clinic ophthalmic technicians who may have variable skills in acquiring OCTA images. Therefore, further evaluation is required in a larger and more diverse cohort of patients including patients with various forms of macular degeneration. The use of trial certified imaging technician may also reduce the rate of unacceptable images in patient cohort. Second, we only examined and categorized the OCTA images based on the en face superficial capillary plexus. The clarity of vascular tree is influenced by segmentation accuracy in addition to motion artifacts. It is commonly noted that segmentation error is more frequent at the outer retinal layers (retinal pigment epithelium and Bruch’s membrane) in macular degeneration. Hence manual grading of the en face superficial capillary plexus may not be applicable to deep avascular layer OCTA image in patients with choroidal neovascularisation. Nevertheless, our manual grading of superficial capillary plexus is more relevant to the disease cohort that we have chosen to study: diabetic retinopathy, vein occlusion and Coats disease. Third, we did not measure the intra individual variation of our manual grading and we did not investigate its impact on the quantitative OCTA measurements and their repeatability. Fourth, the impact of media opacity such as degree of age-related cataract was not taken into account in logistic regression. Our second study also has some potential limitations. First, the small sample size with a heterogeneity of our study patients, including naive patients and treated patients, and eyes with and without macular edema. We were limited in the number of cases with images of acceptable quality for accurately measuring the FAZ and vascular density with several cases were excluded due to poor image quality according to the manual grading system that we developed in our first study.224 Second, we used the automated segmentation to be more reproducible, even though there can be unavoidable mistakes in the segmentation.