Age-related macular degeneration (AMD) is the leading cause of visual impairment in the developed world. Whilst the pathogenesis is complex and not fully understood, changes to the choroidal vasculature in AMD have been demonstrated using histology. Advances in imaging technology, particularly long-wavelength optical coherence tomography (OCT), allow in vivo visualisation and investigation of this structure. The aim of this work is to determine whether changes to the choroidal vasculature are detectable in AMD using in vivo imaging. This was achieved through the evaluation of parameters for quantifying the structure, and the application of a machine learning approach to automated disease severity classification, based on choroidal appearance. Participants with early AMD (n=25), neovascular AMD (nAMD; n=25), and healthy controls (n=25) underwent imaging with a non-commercial long-wavelength (λc=1040 nm) OCT device. Subfoveal choroidal thickness, choroidal area, and luminal area were significantly lower in the nAMD group than the healthy and early AMD groups, whilst vessel ratio was significantly greater (P < 0.05 in all cases). There was no significant difference in visible vessel diameter, choroidal vascularity index, luminal area ratio, or luminal perimeter ratio between the groups. No significant differences were found between the healthy and early AMD groups for any of the eight vascular parameters assessed. Classification of the disease groups based on choroidal OCT images was demonstrated using machine learning techniques. Textural features within the images were extracted using Gabor filters, and K-nearest neighbour, support vector machine, and random forest classifiers were assessed for this classification task. Textural changes were most pronounced in late-stage disease, although attribution to pathology or pharmacological intervention (anti-VEGF treatment) was not possible. Changes were also discernible in the early AMD group, suggesting sensitivity of this approach to detecting vascular involvement in early disease. In conclusion, structural changes to the choroidal vasculature in AMD are detectable in vivo using OCT imaging, demonstrated with both manual and automated analysis techniques. Whilst changes were most prominent in late-stage disease, subtle structural changes in early AMD were identified with texture analysis, warranting further investigation to improve our understanding of choroidal involvement in the pathogenesis of early AMD.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:732243 |
| Date | January 2017 |
| Creators | Terry, Louise |
| Publisher | Cardiff University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://orca.cf.ac.uk/107459/ |
Page generated in 0.0014 seconds