Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets

Abstract

The optic disk (OD) center and margin are typically requisite landmarks in establishing a frame of reference for classifying retinal and optic nerve pathology. Reliable and efficient OD localization and segmentation are important tasks in automatic eye disease screening. This paper presents a new, fast, and fully automatic OD localization and segmentation algorithm developed for retinal disease screening. First, OD location candidates are identified using template matching. The template is designed to adapt to different image resolutions. Then, vessel characteristics (patterns) on the OD are used to determine OD location. Initialized by the detected OD center and estimated OD radius, a fast, hybrid level-set model, which combines region and local gradient information, is applied to the segmentation of the disk boundary. Morphological filtering is used to remove blood vessels and bright regions other than the OD that affect segmentation in the peripapillary region. Optimization of the model parameters and their effect on the model performance are considered. Evaluation was based on 1200 images from the publicly available MESSIDOR database. The OD location methodology succeeded in 1189 out of 1200 images (99% success). The average mean absolute distance between the segmented boundary and the reference standard is 10% of the estimated OD radius for all image sizes. Its efficiency, robustness, and accuracy make the OD localization and segmentation scheme described herein suitable for automatic retinal disease screening in a variety of clinical settings.

Optic disc segmentation using the sliding band filter

“…The parameters of the high-resolution SBF that were set based on the calculated scale factor (α) are shown in the first three rows of Table 4. In order to compare our result with those of the methods [4], [12] and [18], we use the overlapping score (S) that measures the common area between the OD region obtained using the automatic method (A) and the region delimited by experts (E) in [33], being defined by…”

“…The table includes the values for the SBFbased method, the multiresolution ellipse fitting (MEF) [7], the superpixel classification (SPC) approach proposed by Cheng et al [18] and the CHT method [4]. The last row in this table presents the average overlapping score for the the fast and hybrid level set model (F-HLSM) method proposed by Yu et al [12]. As shown in this table, the overlapping between the manually segmented OD region and SBF-based method is always higher than the ones for the other methods, reaching an average overlapping score of 0.89 for the whole set of images in this dataset.…”

“…There are several works on the automatic segmentation of OD in retinal images which can mainly be grouped into four categories, namely template-based methods [4,5,6,7], deformable model methods [8,9,10,11,12,13], morphological-based approaches [14,15,16], and pixel classification methods [17,18]. Within the first category, Aquino et al [4] follow a voting-type algorithm to locate a pixel within the OD as initial information to define a starting sub-image.…”

“…They improved the Chan-Vese model by using local red channel intensities and two texture feature spaces in the neighborhood of the pixels under analysis. The method proposed in [12] uses template matching and a directional matched filter to localize the OD. For OD segmentation, the authors first remove the blood vessels and bright regions using alternating sequential filtering and morphological operations.…”

Automated Detection of New vessels for Classification of Proliferative Diabetic Retinopathy

“…Depending upon the features different channels can be used each contains information of anatomical features e.g. green channel (12) contains information of blood vessels. In this channel vascular structure of vessels can easily be identified and used for processing and red channel contains the information of optic disc.…”