Format

Send to

Choose Destination
See comment in PubMed Commons below
Comput Methods Biomech Biomed Engin. 2013 Apr;16(4):425-34. doi: 10.1080/10255842.2011.623677. Epub 2012 Feb 29.

Automatic detection of age-related macular degeneration pathologies in retinal fundus images.

Author information

1
Department of Biomedical Engineering, Erciyes University, 38039, Kayseri, Turkey. aguven@erciyes.edu.tr

Abstract

Advanced techniques in image processing and analysis are being extensively studied to assist clinical diagnoses. Digital colour retinal fundus images are widely utilised to investigate various eye diseases. In this paper, we describe the detection of optic disc (OD), macula and age-related macular degeneration (ARMD) pathologies of the macular regions in colour fundus images. ARMD causes the loss of central vision in older adults. If the disease is detected early and treated promptly, much of the vision loss can be prevented. Eighty colour retinal fundus images were tested using our proposed algorithm. The Hough transform was employed for OD determination. A fundus coordinate system was established based on the macula location. An ARMD pathology detection methodology using a subtraction process after contrast-limited adaptive histogram equalisation operations was proposed. The accuracies of the automated segmentations of the OD, macula and ARMD pathologies obtained were 100%, 100% and 95.49%, respectively. These results show that our algorithm is a useful tool for detecting ARMD in retinal fundus images. The application of our method may reduce the time needed by ophthalmologists to diagnose ARMD pathology while providing dependable detection precision. Integration of our technique into traditional software could be used in clinical implementations as an aid in disease diagnosis and as a tool for quantitative evaluation of treatment effectiveness.

PMID:
22372623
DOI:
10.1080/10255842.2011.623677
[Indexed for MEDLINE]
PubMed Commons home

PubMed Commons

0 comments
How to join PubMed Commons

    Supplemental Content

    Full text links

    Icon for Taylor & Francis
    Loading ...
    Support Center