Share Email Print

Proceedings Paper

Detection of retinal nerve fiber layer defects in retinal fundus images using Gabor filtering
Author(s): Yoshinori Hayashi; Toshiaki Nakagawa; Yuji Hatanaka; Akira Aoyama; Masakatsu Kakogawa; Takeshi Hara; Hiroshi Fujita; Tetsuya Yamamoto
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

Retinal nerve fiber layer defect (NFLD) is one of the most important findings for the diagnosis of glaucoma reported by ophthalmologists. However, such changes could be overlooked, especially in mass screenings, because ophthalmologists have limited time to search for a number of different changes for the diagnosis of various diseases such as diabetes, hypertension and glaucoma. Therefore, the use of a computer-aided detection (CAD) system can improve the results of diagnosis. In this work, a technique for the detection of NFLDs in retinal fundus images is proposed. In the preprocessing step, blood vessels are "erased" from the original retinal fundus image by using morphological filtering. The preprocessed image is then transformed into a rectangular array. NFLD regions are observed as vertical dark bands in the transformed image. Gabor filtering is then applied to enhance the vertical dark bands. False positives (FPs) are reduced by a rule-based method which uses the information of the location and the width of each candidate region. The detected regions are back-transformed into the original configuration. In this preliminary study, 71% of NFLD regions are detected with average number of FPs of 3.2 per image. In conclusion, we have developed a technique for the detection of NFLDs in retinal fundus images. Promising results have been obtained in this initial study.

Paper Details

Date Published: 30 March 2007
PDF: 8 pages
Proc. SPIE 6514, Medical Imaging 2007: Computer-Aided Diagnosis, 65142Z (30 March 2007); doi: 10.1117/12.710181
Show Author Affiliations
Yoshinori Hayashi, Graduate School of Medicine, Gifu Univ. (Japan)
TAK Co., Ltd. (Japan)
Toshiaki Nakagawa, Graduate School of Medicine, Gifu Univ. (Japan)
Yuji Hatanaka, Gifu National College of Technology (Japan)
Akira Aoyama, Graduate School of Medicine, Gifu Univ. (Japan)
Masakatsu Kakogawa, TAK Co., Ltd. (Japan)
Takeshi Hara, Graduate School of Medicine, Gifu Univ. (Japan)
Hiroshi Fujita, Graduate School of Medicine, Gifu Univ. (Japan)
Tetsuya Yamamoto, Graduate School of Medicine, Gifu Univ. (Japan)

Published in SPIE Proceedings Vol. 6514:
Medical Imaging 2007: Computer-Aided Diagnosis
Maryellen L. Giger; Nico Karssemeijer, Editor(s)

© SPIE. Terms of Use
Back to Top