No abstract

The application of image processing to the investigation of Age-related Macular Degeneration (AMD) has focused on detecting focal drusen deposits in colour fundus images. This research investigates Gradient Vector Flow (GVF) active contours for the detection of choroidal neovascularisation (CNV) from fundus fluorescein angiograms in exudative AMD, the most severe form of the disease. The method was used in the identification of hyperfluorescent regions using pre-selected angiograms with expert-identified lesion components. Using active contours that are positioned close enough to the lesion of interest by the user, the algorithm can simply deform to the lesion, providing a successful outcome. If the initial active contour is positioned too far from the lesion location, the algorithm had a tendency to identify other image components incorrectly, especially in images with significant background interference or complex lesions. The algorithm reported here is guided by experts and hence is semi-automatic. The solutions (lesion position and size) obtained were compared with those identified and measured by an expert reader in a series of 10 fundus fluorescein images.

Purpose In 2004, the Verteporfin Photodynamic Therapy (VPDT) study was commissioned in the UK to assess patients with CNV undergoing PDT. A Reading Centre network (NetwORC UK) was established to independently grade 5000+ angiograms per year. Previous to this, there were no independent RCs in Europe capable of grading this volume. Methods Three geographically distinct centres (Belfast, Liverpool and London) form NetwORC UK with a management facility (CARF) coordinating the administrative and technical aspects. A custom software platform was developed to manage the import, transmission and grading of images, collection of data between the three RCs, and ensure communication across separate IT networks. Quality assurance and concordance ensure grading is consistent across NetwORC UK. Results Between October 2005 and October 2007, 12612 angiograms were graded and 6833 were baseline angiograms. The grading of angiograms is ongoing in NetwORC UK and will continue until the study ceases. Conclusion A network structure has enabled grading of a large volume of images from the VPDT study. Regular training and concordance exercises have contributed to the success of this service. The authors recommend that the administration and management is separate from grading ensuring grading staff are dedicated to grading tasks. Therefore technical queries and software can be managed in one location, importing and exporting data can be centralised, vacation and sickness cover is easier to maintain, and unique training is offered by pooling staff from different locations. Regular monitoring of IT systems ensures data has integrity and dedicated IT personnel are essential. Future studies of this size will now be possible in Europe using NetwORC UK.

Purpose To report on concordance of optical coherence tomography (OCT) grading in the Network of Ophthalmic Reading Centres UK (NetwORC UK). The present study looks at concordance of OCT gradings across the network for patients enrolled in the UK IVAN Study. Methods A set of 5 patient scans (6 radial line scans and one 7mm offset scan using the Stratus OCT III (Zeiss, UK)) were graded by accredited graders at all three centres. Scans were exported as JPEG images with a corresponding PDF. Measurements for the Outer High Reflectivity Band (OHRB), Sub Retinal Fluid (SRF), Pigment Epithelial Detachment (PED), maximum height of largest cyst, maximum retinal thickenss (MRT) and foveal retinal thickness (FRT) were taken, if present. Results In all cases an acceptable level of concordance was achieved. One case proved more problematic than the others and results were more varied. For OHRB, SRF, PED, cyst, MRT and FRT the measurements by individual graders were consistent in 4 out of 5 cases. One case showed wider‐ranging measurements with the range for OHRB varying by 0.25mm (mean thickness 0.22mm). Conclusion The present study shows an acceptable level of concordance for OCT grading across NetwORC UK. In difficult cases, the complexity of AMD grading explained the lack of total concordance when an abnormal thickening of the OHRB is present. The most problematic case had an area of moderate hyperreflectivity anterior to the OHRB which caused difficulty and highlighted a specific area for further training and a protocol amendment. The exercise was useful in highlighting both individual and overall training needs which are currently being addressed in each of the RCs. The concordance study is being extended to include a larger dataset.

Purpose: To report on the identification of Retinal Angiomatous Proliferation (RAP) in the Network of Ophthalmic Reading Centres UK (NetwORC) for the Verteporfin Photodynamic Therapy (VPDT) cohort study. RAP tends to be bilateral and although it is accepted to be part of the spectrum of age‐related macular degeneration (AMD), no treatment has yet proved to be as successful as in other types of AMD. A diagnosis of RAP is therefore instrumental to the success of treatment decisions to be made. Methods: The VPDT study involves the grading of more than 5000 angiograms per annum at the network of Reading Centres (RC) with regular training and concordance activities being performed. The grading database was queried for RAP lesions identified on a monthly basis between Nov 2005 and Oct 2006. Expert training on identification of RAP was given at 3 successive meetings during this period. Results: In this 12‐month period, 5402 angiograms were graded (including both primary and quality assurance gradings). RAP was identified in 264 cases, with a step‐wise increase in the identification of RAP during this one year period. Using a simple chi‐squared for trend test, there is a highly significant increasing trend over the 12 months (p<0.0001) following training sessions at all three RCs, highlighting the benefit of structured training. Conclusions: The present study has shown that consistent concordance and training exercises across the RC network improves the identification of specific lesion components such as RAP. With improved education, other clinical and grading staff can be trained to recognise RAP lesions and better identification may lead to improved treatment outcomes.