What is the radial displacement of the peak gradient?

The normalized negative gaussian-smoothed gradient magnitude image, Γ = −‖∇I‖/ max(‖∇I‖), is sampled to produce radial gradient profiles; the radial displacement of the “peak gradient” within a limited search range about mi, denoted gi, provides an external attractor point.

What is the phase two of the global model?

Phase two roughly locates the rim, by locking the global model onto the relatively reliable and distractor-free temporal rim; phase three fits it to the entire rim.

What is the way to locate the optic nerve head?

In fact the single filter used suffices; where the size mismatches the maximal correlation lies on an annulus around the optic nerve head center; the peak istherefore still located within the optic nerve head, which is sufficient for their requirements.

What did the authors do to make the Hough search more reliable?

The authors looked for ellipses of width 56–112, aspect ratio 1.03, and smoothed the Hough-space with a Gaussian kernel (σ = 1.5) before peak detection to account for deformations.

What is the method to reduce distractors?

A method suggested by both Lee [22] and Mendels [12] to reduce distractors is to morphologically remove the vessels; the authors refer to this as “de-vascularization”, and implement it by grey-scale morphological closure with a disk-shaped structuring element, radius 7.

What is the gradient of Eext with respect to r?

The gradient of Eext with respect to r is given by equation 31, derived from equation 21, bearing in mind that to enforce the global model, the authors set mi = air.

What is the preferred method for grading ocular images?

An expert human screener grades the images; this is a time-consuming, highly skilled task, susceptible to subjective variation and error.

What is the difference between freeform and parametric snakes?

Freeform snakes provide many local degrees of freedom, whereas parametric deformable models encode a specific shape and its variations [18] [19]; the model can help to overcome problems in the images such as boundary gaps.

What is the role of the optic nerve head in the retina?

the optic nerve head is important in localization of the fovea, the central part of the retina that subserves fine vision.

What is the heuristic method for determining the distance from the fovea?

The “2.5 disk diameters” heuristic method is therefore more practical for their application, particularly as the authors do not require absolute accuracy.

How many iterations of the optimization are necessary?

To compensate for this, the authors iteratively recalculate the aspect ratio from the attractor points at the end of the phase, and repeat the optimization (which is extremely fast); three iterations suffice.

Open AccessJournal Article10.1109/TMI.2003.823261

Optic nerve head segmentation

J. Lowell, +6 more

- 06 Feb 2004

- IEEE Transactions on Medical Imaging

- Vol. 23, Iss: 2, pp 256-264

452

TL;DR: The localization algorithm succeeded on all bar one usable image; the contour estimation algorithm was qualitatively assessed by an ophthalmologist as having Excellent-Fair performance in 83% of cases, and performs well even on blurred images.

Abstract: Reliable and efficient optic disk localization and segmentation are important tasks in automated retinal screening. General-purpose edge detection algorithms often fail to segment the optic disk due to fuzzy boundaries, inconsistent image contrast or missing edge features. This paper presents an algorithm for the localization and segmentation of the optic nerve head boundary in low-resolution images (about 20 /spl mu//pixel). Optic disk localization is achieved using specialized template matching, and segmentation by a deformable contour model. The latter uses a global elliptical model and a local deformable model with variable edge-strength dependent stiffness. The algorithm is evaluated against a randomly selected database of 100 images from a diabetic screening programme. Ten images were classified as unusable; the others were of variable quality. The localization algorithm succeeded on all bar one usable image; the contour estimation algorithm was qualitatively assessed by an ophthalmologist as having Excellent-Fair performance in 83% of cases, and performs well even on blurred images.

AI Agents for this Paper

Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps

Most frequently asked questions

1. What contributions have the authors mentioned in the paper "Optic nerve head segmentation" ?

This paper presents an algorithm for the localization and segmentation of the optic nerve head boundary in low-resolution images ( about 20μ per pixel ).

2. What future works have the authors mentioned in the paper "Optic nerve head segmentation" ?

Future work to improve the algorithm to deal with these distractors will require reasoning at a higher level.. Although their current approach focuses on detecting the optic nerve head boundary as a stage in the structured analysis of images for diabetic retinopathy fundus images, the approach is also of relevance in diagnosis of other diseases, and particularly in screening for glaucoma ; in the future the authors will evaluate the method for use in detection of the rim in glaucomatous images.

3. How can the authors transform an elliptical model into a circular one?

The elliptical model can be transformed into an equivalent circular model by scaling using the spoke ratios, ai (equation 11), to “normalize” radial distances and forces.

4. What is the main argument for a structured approach to grading retinal images?

The variability of retinal appearance argues for a structured approach [5], where significant landmarks in the retina are identified, and the relationships between them exploited to increase confidence in the classification of each object.

TABLE I Visual Characteristics. Characteristics that present potential problems are enumerated, in three categories: problems rendering images poor enough to exclude; affecting localization; and affecting segmentation. Some images have multiple characteristics.

Fig. 6. Sample Segmentations. a) Excellent. b) Good. c) Fair. d) Poor. Solid line: algorithm; dotted line: mean clinician boundary.

TABLE III Subjective Classification of Performance

Citations

•Journal Article•10.1109/TMI.2019.2903562

CE-Net: Context Encoder Network for 2D Medical Image Segmentation

Zaiwang Gu, +8 more

- 07 Mar 2019

- IEEE Transactions on Medical Imaging

TL;DR: Comprehensive results show that the proposed CE-Net method outperforms the original U- net method and other state-of-the-art methods for optic disc segmentation, vessel detection, lung segmentation , cell contour segmentation and retinal optical coherence tomography layer segmentation.

...read moreread less

1.4K

•Journal Article•10.1109/TMI.2019.2903562

CE-Net: Context Encoder Network for 2D Medical Image Segmentation

Zaiwang Gu, +8 more

- 07 Mar 2019

- arXiv: Computer Vision and Pattern Recog...

TL;DR: Li et al. as mentioned in this paper proposed a context encoder network (referred to as CE-Net) to capture more high-level information and preserve spatial information for 2D medical image segmentation, which mainly contains three major components: a feature encoder module, a context extractor and a feature decoder module.

...read moreread less

1.1K

•Journal Article•10.1109/TMI.2018.2791488

Joint Optic Disc and Cup Segmentation Based on Multi-Label Deep Network and Polar Transformation

Huazhu Fu, +5 more

- 09 Jan 2018

- IEEE Transactions on Medical Imaging

TL;DR: Zhang et al. as discussed by the authors proposed a multi-scale input layer, U-shape convolutional network, side-output layer, and multi-label loss function for OD and OC segmentation.

...read moreread less

942

•Journal Article•10.1016/J.MEDIA.2019.101570

REFUGE Challenge: A unified framework for evaluating automated methods for glaucoma assessment from fundus photographs

José Ignacio Orlando, +30 more

- 01 Jan 2020

- Medical Image Analysis

TL;DR: It is observed that two of the top-ranked teams outperformed two human experts in the glaucoma classification task, and the segmentation results were in general consistent with the ground truth annotations, with complementary outcomes that can be further exploited by ensembling the results.

...read moreread less

708

Journal Article•10.1016/J.PRETEYERES.2005.07.001

Retinal image analysis: Concepts, applications and potential

Niall Patton, +8 more

- 01 Jan 2006

- Progress in Retinal and Eye Research

TL;DR: The use of image analysis in the automated diagnosis of pathology (with particular reference to diabetic retinopathy) is reviewed, as well as its role in defining and performing quantitative measurements of vascular topography, and how these entities are based on 'optimisation' principles.

...read moreread less

658

...

Expand

References

Numerical recipes in C

William H. Press, +3 more

- 01 Jan 1994

TL;DR: The Diskette v 2.06, 3.5''[1.44M] for IBM PC, PS/2 and compatibles [DOS] Reference Record created on 2004-09-07, modified on 2016-08-08.

...read moreread less

20.7K

•Book

Neural networks for pattern recognition

Christopher M. Bishop

- 01 Jan 1995

TL;DR: This is the first comprehensive treatment of feed-forward neural networks from the perspective of statistical pattern recognition, and is designed as a text, with over 100 exercises, to benefit anyone involved in the fields of neural computation and pattern recognition.

...read moreread less

19.9K

Journal Article•10.1007/BF00133570

Snakes : Active Contour Models

Michael Kass, +2 more

- 01 Jan 1988

- International Journal of Computer Vision

TL;DR: This work uses snakes for interactive interpretation, in which user-imposed constraint forces guide the snake near features of interest, and uses scale-space continuation to enlarge the capture region surrounding a feature.

...read moreread less

19.1K

Book Chapter•10.1093/oso/9780198538493.001.0001

Neural Networks for Pattern Recognition

Christopher M. Bishop

- 23 Nov 1995

Abstract: Abstract This book provides the first comprehensive treatment of feed-forward neural networks from the perspective of statistical pattern recognition. After introducing the basic concepts of pattern recognition, the book describes techniques for modelling probability density functions, and discusses the properties and relative merits of the multi-layer perceptron and radial basis function network models. It also motivates the use of various forms of error functions, and reviews the principal algorithms for error function minimization. As well as providing a detailed discussion of learning and generalization in neural networks, the book also covers the important topics of data processing, feature extraction, and prior knowledge. The book concludes with an extensive treatment of Bayesian techniques and their applications to neural networks.

...read moreread less

9.8K

Journal Article•10.1109/42.845178

Locating blood vessels in retinal images by piecewise threshold probing of a matched filter response

A.D. Hoover, +2 more

- 01 Mar 2000

- IEEE Transactions on Medical Imaging

TL;DR: An automated method to locate and outline blood vessels in images of the ocular fundus that uses local and global vessel features cooperatively to segment the vessel network is described.

...read moreread less

2.6K

...

Expand

Optic nerve head segmentation

Chat with Paper

AI Agents for this Paper

Most frequently asked questions

1. What contributions have the authors mentioned in the paper "Optic nerve head segmentation" ?

2. What future works have the authors mentioned in the paper "Optic nerve head segmentation" ?

3. How can the authors transform an elliptical model into a circular one?

4. What is the main argument for a structured approach to grading retinal images?

5. What is the radial displacement of the peak gradient?

6. What is the phase two of the global model?

7. What is the way to locate the optic nerve head?

8. What did the authors do to make the Hough search more reliable?

9. What is the method to reduce distractors?

10. What is the gradient of Eext with respect to r?

11. What is the preferred method for grading ocular images?

12. What is the difference between freeform and parametric snakes?

13. What is the role of the optic nerve head in the retina?

14. What is the heuristic method for determining the distance from the fovea?

15. How many iterations of the optimization are necessary?

Figures

Citations

CE-Net: Context Encoder Network for 2D Medical Image Segmentation

CE-Net: Context Encoder Network for 2D Medical Image Segmentation

Joint Optic Disc and Cup Segmentation Based on Multi-Label Deep Network and Polar Transformation

REFUGE Challenge: A unified framework for evaluating automated methods for glaucoma assessment from fundus photographs

Retinal image analysis: Concepts, applications and potential

References

Numerical recipes in C

Neural networks for pattern recognition

Snakes : Active Contour Models

Neural Networks for Pattern Recognition

Locating blood vessels in retinal images by piecewise threshold probing of a matched filter response

Related Papers (5)

Locating the optic nerve in a retinal image using the fuzzy convergence of the blood vessels

Detecting the Optic Disc Boundary in Digital Fundus Images Using Morphological, Edge Detection, and Feature Extraction Techniques

Automated localisation of the optic disc, fovea, and retinal blood vessels from digital colour fundus images.

Detection of optic disc in retinal images by means of a geometrical model of vessel structure

Ridge-based vessel segmentation in color images of the retina