• 作者：Antonio Pose-Reino^a ; ^{antoniopose@telefonica.net} ; Marta Pena Seijo^a ; Manuel Gonzá ; lez Penedo^b ; Marcos Ortega Hortas^b ; Marí ; a Rodrí ; guez Blanco^c ; Pilar Vega^d ; José ; Luis Dí ; az Dí ; az^d ; Noa Ferná ; ndez^e ; Juan Carlos Esté ; vez^f ; Francisco Gó ; mez-Ulla Irazá ; bal^c
• 关键词：Retinopatí ; a hipertensiva ; Í ; ndice arteriovenoso ; Mé ; todo semiautom&aacute ; tico
• 刊名：Medicina Cl铆nica
• 出版年：2010
• 期刊代码：pca254_00257753
• 类别：med
• 出版时间：3 July 2010
• 卷：135
• 期：4
• 页码：145-150
• 文件大小：354 K

Background and objective

Early alterations in retinal microcirculation are observed in most hypertensive patients seen in daily practice and the European guidelines consider it is necessary an objective method to quantify these alterations. We have previously described a semi-automatic computerized system to evaluate the calibre of retinal blood vessels that has shown high sensitivity and specificity to calculate the arteriovenous ratio (AVR), though with limitations. We describe a method based on the snakes model to calculate the arteriovenous ratio.

Patients and method

We haved compared it with the previously reported lineal method, and we have analyed 173 digital photographs from 96 hypertensive patients, most of them reciveing treatment. Photos were made in two hospitals (A Coru帽a: 66 and Santiago de Compostela: 107), and were analysed by the same people in each Centre, by the lineal and snake method.

Results

We have observed that the arithmetic mean and the differences in AVR between both methods were minimal and showed a normal distribution. Cronbach statistics was 0.974 and intraclass correlation coefficient 0.949 (p<0.001), for the images from Santiago and 0.923 with an intraclass correlation coefficient of 0.857 (p<0.001) for the images from A Coru帽a.

Conclusion

This semiautomatic method to calculate the AVR ratio has a high sensitivity and a greater specificity than previous method, and the correlation between the results obtained with both is excellent.