Purchase this article with an account.
Jeffrey C Wigdahl, Pedro Guimaraes, Enea Poletti, Alfredo Ruggeri; ReVMS (Retinal Vasculature Measurement System). Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):5260.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
This work provides ophthalmologists with a semi-automatic tool for the complete analysis of the retinal vasculature. It will allow clinicians and researchers a quantitative assessment of clinical vessel parameters, provide a framework to test new automatic algorithms, and have the ability to edit results at several steps in process.
The application combines automatic algorithms for vessel segmentation, optic disc detection, and Artery/Vein classification, with the ability to edit vessel segments, for the calculation of common clinical vessel parameters (tortuosity, artery/vein ratio (AVR), fractal dimension, AV nicking). After vessel segmentation, the user can edit vessel segments (add new, remove, combine, split). Vessel widths, tortuosity metrics, and fractal dimension are calculated. Optic disc center and radius are found and can be manually adjusted.<br /> Artery/Vein classification is performed in an area between 1 and 2 disc diameters from the optic disc. The user has the ability to edit, add/remove vessels, and change vessel widths before automatic AV classification. The user can edit results again before AV ratio, central retinal vein equivalent, and central retinal artery equivalent calculation. Finally, vessel crossover points are analyzed to detect narrowing and angle in/out. Vessel widths can be edited prior to calculation. All clinical parameters are computed using our previously developed methods. Tortuosity is measured per segment and combined to give a whole image tortuosity. The set of 45 images used to test the system are from the High Resolution Fundus (HRF) image database provided by Friedrich-Alexander University Erlangen-Nuremberg. Images are 2336x3504 pixels covering a 50 degree field of view.
Intermediate steps of the system can be seen in Fig. 1. The system has been used to detect vessels, compute tortuosity and AV ratio in a set of 45 high quality images.
This system provides a framework for the quantitative analysis of the retinal vasculature with the intention of making the automatic portions interchangeable, for the testing of new algorithms, while maintaining editing functions, to ensure the highest quality results. Speed, ease of use, and clinical effectiveness were the main criteria in system development.
This PDF is available to Subscribers Only