Loading…

Decision support system for detection of hypertensive retinopathy using arteriovenous ratio

•Accurate detection of Hypertensive Retinopathy (HR) and classification of moderate and malignant HR.•Lower number of false positives by using accurate blood vessel and A/V(Artery/Vein) classification.•Color statistical features based vector for A/V classification.•Achieved high accuracies for class...

Full description

Saved in:
Bibliographic Details
Published in:Artificial intelligence in medicine 2018-08, Vol.90, p.15-24
Main Authors: Akbar, Shahzad, Akram, Muhammad Usman, Sharif, Muhammad, Tariq, Anam, Khan, Shoab A.
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Accurate detection of Hypertensive Retinopathy (HR) and classification of moderate and malignant HR.•Lower number of false positives by using accurate blood vessel and A/V(Artery/Vein) classification.•Color statistical features based vector for A/V classification.•Achieved high accuracies for classification of moderate and malignant HR. Hypertensive Retinopathy (HR) caused by hypertension is a retinal disease which may leads to vision loss and blindness. Computer aided diagnostic systems for various diseases are being used in clinics but there is a need to develop an automated system that detects and grades HR disease. In this paper, an automated system is presented that detects and grades HR disease using Arteriovenous Ratio (AVR).The presented system includes three modules i.e. main component extraction, artery/vein (A/V) classification and finally AVR calculation and grading of HR. Proposed system uses vascular map and a set of hybrid features for A/V classification. The evaluation of proposed system is carried out using three datasets. The proposed system shows average accuracies of 95.14% for images of INSPIRE-AVR database, 96.82% for images of VICAVR database and 98.76% for local dataset AVRDB. These results support that the proposed system is trustworthy for clinical use in detection and grading of HR disease. Main contribution of proposed system is that it utilizes complete blood vessel map for A/V classification. These arteries and veins are then used to calculate AVR and grade HR cases based on AVR values. Another contribution of this article is that it presents a new dataset AVRDB for A/V classification and HR detection.
ISSN:0933-3657
1873-2860
DOI:10.1016/j.artmed.2018.06.004