Quantification of blood vessel calibre in retinal images of multi-ethnic school children using a model based approach.
Comput Med Imaging Graph
; 37(1): 48-60, 2013 Jan.
Article
em En
| MEDLINE
| ID: mdl-23410507
ABSTRACT
Changes and variation in retinal vessel width are related to vascular risk factors and prospectively related to cardiovascular disease in later life. Hence, assessment of vessel width may be a useful physio-marker and potential predictor of cardiovascular status. However, measurement of vessel calibre from retinal images is a challenging process to automate. This paper proposes an automated system to measure vessel calibre in retinal images, which is demonstrated in images of multi-ethnic school children. The diameter measurement is based on the detection of the centreline pixels from a vessel probability map image, determining the vessel orientation at these pixels, extracting the vessel segments and later using a two-dimensional model, which is optimized to fit various types of intensity profiles of vessel segments. The width is then estimated from parameters of the optimized model. The method is also quantitatively analyzed using monochromatic representations of different colour spaces. The algorithm is evaluated on a recently introduced public database CHASE_DB1, which is a subset of retinal images of multi-ethnic children from the Child Heart and Health Study in England (CHASE) dataset. Moreover, the precise estimation of retinal vascular widths is critical for epidemiologists to identify the risk factors. This work also introduces an interactive software tool for epidemiologists, with which retinal vessel calibre can be precisely marked.
Texto completo:
1
Coleções:
01-internacional
Base de dados:
MEDLINE
Assunto principal:
Oftalmoscopia
/
Vasos Retinianos
/
Reconhecimento Automatizado de Padrão
/
Etnicidade
Tipo de estudo:
Etiology_studies
/
Prognostic_studies
/
Risk_factors_studies
Limite:
Child
/
Female
/
Humans
/
Male
País/Região como assunto:
Europa
Idioma:
En
Ano de publicação:
2013
Tipo de documento:
Article