---

Working Fields

Retinal Image Analysis

One of the most active fields of work in the VARPA Group is the ophtalmology, in particular the analysis of eye fundus images (retinal images). These images are mostly granted by the Complejo Hospitalario Universitario of Santiago (CHUS).

The retinal image processing is a very interesting and demanding field, having a lot of practical applications, such as the development of applications for massive medical revision and the research in pharmacology effectiveness.

Some of the most important active topics are:

Drusen detection

Drusen are tiny white or yellow spots associated with the age-related macular degeneration (AMD). This disease can lead to severe loss central vision and adversely affect the patient's quality of life. We have developed an automatic methodology to detect drusen in initial stages. This methodology is based a template matching technique to find the drusen in the interest areas. Currently, our work is focused on the analysis of the drusen evolution in OCTR sequences.

Red lesion detection

The diabetic retinopathy is a disease related to the diabetes that can cause blindness. One important symptom of diabetic retinopathy is the development of red lesions in the retina. We have developed an automatic methodology to analyse the retinal images in order to find this red lesions. This methodology is based on correlation filters and region growing segmentation techniques.

Automatic computation of the Arteriolar-to-Venular ratio

The retina Arteriolar-to-Venular ratio (AVR) is a parameter that helps the diagnosis of some pathologies, such as hypertension or arteriosclerosis. It is mainly computed as the ratio between the sum of artery calibers and the sum of vein calibers in several circumferences centered at the optic disc. We are currently developing and testing an automatic methodology to compute the AVR. This methodology involves the following steps:

1. Optic disc location. It is the starting point. We are testing several algorithms to perform an efficient and reliable detection to the optic disc.
2. Selection of Interest Radii centered at the optic disc.
3. Vessel Detection and Caliber Measurement. We have developed a methodology based on snakes to measure the vessel diameter in concentric circumferences.
4. Vessel Classification into Arteries and Veins. This is the most difficult step in the whole process due to the variability within inter-patient and intra-patient vessel contrast. We are testing several techniques and classifiers in order to minimize the misclassifications.
5. AVR Computation. In the final step, we compute the AVR using the information obtained in the previous steps.

Applications

Our goals are the development of user-friendly applications for support diagnosis and disease monitoring. In this sense, we are working in the development of a complete application for the automatic computation of the Arteriolar-to-Venular ratio, the SIRIUS web application. The use of web technologies makes easier the software maintenance and the collaboration of several experts from different medical institutions.

Url: http://www.varpa.org/Sirius

Public databases

The VICAVR database is a set of retinal images used for the computation of the A/V Ratio. The database currently includes 58 images. The images have been acquired with a TopCon non-mydriatic camera NW-100 model and are optic disc centered with a resolution of 768x584. The database includes the caliber of the vessels measured at different radii from the optic disc as well as the vessel type (artery/vein) labelled by three experts.

Main publications

• C. Alonso-Montes, D. L. Vilarino, P. Dudek, M. G. Penedo, "Fast Retinal Vessel Tree Extraction: A Pixel Parallel Approach", International Journal of Circuit Theory and Application (Special Issue Cellular Wave Computing Architectures), 36, 641-651, 2008.
  [Abstract] [PDF] [+Info.]
  Early ocular disease diagnosis has became one of the main fields in medical research. From the image processing point of view, many strategies and algorithms have been developed to deal with the extraction of the retinal vessel tree. Although reliable and accurate results have been obtained, the main disadvantage in most of these proposals is still the high computation effort required. In this paper, a methodology to extract the retinal vessel tree has been developed and tested in a fine-grain pixel-parallel processor array. An analysis of the execution time has been made to show its capabilities regarding the computation speed. Moreover, an analysis of the accuracy using a publicly available database has been made to show the algorithm performance.
• J. Novo, M. G. Penedo, J. Santos, "Optic disc segmentation by means of GA-Optimized Topological Active Nets", Lecture Notes in Computer Science: Image Analysis and Recognition., 5112, 807-816, 2008.
  [Abstract] [PDF] [+Info.]
  In this paper we propose a new approach to the optic disc segmentation process in digital retinal images by means of Topological Active Nets (TAN). This is a deformable model used for image segmentation that integrates features of region-based and edge-based segmentation techniques, being able to fit the edges of the objects and model their inner topology. The optimization of the Active Nets is performed by a genetic algorithm, with adapted or new ad hoc genetic operators to the problem. The active nets incorporate new energy terms for the optic disc segmentations, without the need of any pre-processing of the images. We present results of optic disc segmentations showing the advantages of the approach.
• A. Pose-Reino, F. Gómez-Ulla, B. Hayik, M. J. Carreira, A. Mosquera, M. G. Penedo, "Computerized measuremente of retinal blood vessel calibre: description, validation and use to determine the influence of ageing and hypertension", Journal of hypertensión, 23(4), 843-50, 2005.
  [+Info.]