1

CHAPTER -1 INTRODUCTION

1. INTRODUCTION

1.1 IMAGE SEGMENTATION

Segmentation refers to the process of partitioning a digital image into multiple segments (sets of pixels, also known as superpixels). The goal of segmentation is to simplify and/or change the representation of an image into something that is more meaningful and easier to analyze. Image segmentation is typically used to locate objects and boundaries (lines, curves, etc.) in images. More precisely, image segmentation is the process of assigning a label to every pixel in an image such that pixels with the same label share certain visual characteristics.

1.2 APPLICATIONS Some of the practical applications of image segmentation are:

y

Medical imaging
o o o o o o

Locate tumors and other pathologies Measure tissue volumes Computer-guided surgery Diagnosis Treatment planning Study of anatomical structure

y y y

Locate objects in satellite images (roads, forests, etc.) Face recognition Iris recognition

2 Fingerprint recognition Traffic control systems Brake light detection Machine vision Agricultural imaging crop disease detection

y y y y y

Several general-purpose algorithms and techniques have been developed for image segmentation. Since there is no general solution to the image segmentation problem, these techniques often have to be combined with domain knowledge in order to effectively solve an image segmentation problem for a problem domain.

1.3 VARIOUS SEGMENTATION METHODS

There exist various types of image segmentation techniques 1.3.1 THRESHOLDING The simplest method of image segmentation is called the thresholding method. This method is based on a clip-level (or a threshold value) to turn a gray-scale image into a binary image. The key of this method is to select the threshold value. Several popular methods are used in industry including the maximum entropy method, Otsu's method (maximum variance), and k-means clustering can also be used. 1.3.2 COMPRESSION-BASED METHODS Compression based methods postulate that the optimal segmentation is the one that minimizes, over all possible segmentations, the coding length of the data. The connection between these two concepts is that segmentation tries to find patterns in an image and any regularity in the image can be used to compress it. The method describes each segment by its texture and boundary shape. Each of these components is modeled by a probability distribution function and its coding length is computed as follows:

3 1. The boundary encoding leverages the fact that regions in natural images tend to have a smooth contour. This prior is used by Huffman coding to encode the difference chain code of the contours in an image. 2. Texture is encoded by lossy compression in a way similar to minimum description length (MDL) principle. For any given segmentation of an image, this scheme yields the number of bits required to encode that image based on the given segmentation. Thus, among all possible segmentations of an image, the goal is to find the segmentation which produces the shortest coding length. This can be achieved by a simple agglomerative clustering method. 1.3.3 HISTOGRAM-BASED METHODS Histogram-based methods are very efficient when compared to other image segmentation methods because they typically require only one pass through the pixels. In this technique, a histogram is computed from all of the pixels in the image, and the peaks and valleys in the histogram are used to locate the clusters in the image. Color or intensity can be used as the measure. A refinement of this technique is to recursively apply the histogram-seeking method to clusters in the image in order to divide them into smaller clusters. This is repeated with smaller and smaller clusters until no more clusters are formed. One disadvantage of the histogram-seeking method is that it may be difficult to identify significant peaks and valleys in the image. In this technique of image classification distance metric and integrated region matching are familiar. Histogram-based approaches can also be quickly adapted to occur over multiple frames, while maintaining their single pass efficiency. The histogram can be done in multiple fashions when multiple frames are considered. The same approach that is taken with one frame can be applied to multiple, and after the results are merged, peaks and valleys that were previously difficult to identify are more likely to be distinguishable. The

4 histogram can also be applied on a per pixel basis where the information results are used to determine the most frequent color for the pixel location. This approach segments based on active objects and a static environment, resulting in a different type of segmentation useful in Video tracking. 1.3.4 EDGE DETECTION Edge detection is a well-developed field on its own within image processing. Region boundaries and edges are closely related, since there is often a sharp adjustment in intensity at the region boundaries. Edge detection techniques have therefore been used as the base of another segmentation technique. The edges identified by edge detection are often disconnected. To segment an object from an image however, one needs closed region boundaries. Edge is nothing but boundry between two images. Edge detection technique refers to the identification and locating the sharp discontinuities in the image. 1.3.5 REGION GROWING METHODS The first region growing method was the seeded region growing method. This method takes a set of seeds as input along with the image. The seeds mark each of the objects to be segmented. The regions are iteratively grown by comparing all unallocated neighbouring pixels to the regions. The difference between a pixel's intensity value and the region's mean, , is used as a measure of similarity. The pixel with the smallest difference measured this way is allocated to the respective region. This process continues until all pixels are allocated to a region. Seeded region growing requires seeds as additional input. The segmentation results are dependent on the choice of seeds. Noise in the image can cause the seeds to be poorly placed. Unseeded region growing is a modified algorithm that doesn't require explicit seeds. It starts off with a single region A1 the pixel chosen here does not significantly influence final segmentation. At each iteration it considers the neighbouring pixels in the same way as seeded region growing. It differs from seeded region growing

5 in that if the minimum is less than a predefined threshold T then it is added to the

respective region Aj. If not, then the pixel is considered significantly different from all current regions Ai and a new region An + 1 is created with this pixel. One variant of this technique is based on pixel intensities. The mean and scatter of the region and the intensity of the candidate pixel is used to compute a test statistic. If the test statistic is sufficiently small, the pixel is added to the region, and the regions mean and scatter are recomputed. Otherwise, the pixel is rejected, and is used to form a new region. A special region growing method is called -connected segmentation. It is based on pixel intensities and neighborhood linking paths. A degree of connectivity will be calculated based on a path that is formed by pixels. For a certain value of , two pixels are called -connected if there is a path linking those two pixels and the connectedness of this path is at least . -connectedness is an equivalence relation. 1.3.6 SPLIT-AND-MERGE METHODS Split-and-merge segmentation is based on a quad tree partition of an image. It is sometimes called quad tree segmentation. This method starts at the root of the tree that represents the whole image. If it is found non-uniform, then it is split into four son-squares (the splitting process), and so on so forth. Conversely, if four son-squares are homogeneous, they can be merged as several connected components (the merging process). The node in the tree is a segmented node. This process continues recursively until no further splits or merges are possible. When a special data structure is involved in the implementation of the algorithm of the method, its time complexity can reach O(nlogn), an optimal algorithm of the method. 1.3.7 PARTIAL DIFFERENTIAL EQUATION-BASED METHODS Using a partial differential equation (PDE)-based method and solving the PDE equation by a numerical scheme, one can segment the image.

1.3.8 GRAPH PARTITIONING METHODS Graph partitioning methods can effectively be used for image segmentation. In these methods, the image is modeled as a weighted, undirected graph. Usually a pixel or a group of pixels are associated with nodes and edge weights define the (dis)similarity between the neighborhood pixels. The graph (image) is then partitioned according to a criterion designed to model "good" clusters. Each partition of the nodes (pixels) output from these algorithms are considered an object segment in the image. Some popular algorithms of this category are normalized cuts, random walker, minimum cut, isoperimetric partitioning and minimum spanning tree-based segmentation. 1.3.9 WATERSHED TRANSFORMATION The watershed transformation considers the gradient magnitude of an image as a topographic surface. Pixels having the highest gradient magnitude intensities (GMIs) correspond to watershed lines, which represent the region boundaries. Water placed on any pixel enclosed by a common watershed line flows downhill to a common local intensity minimum (LIM). Pixels draining to a common minimum form a catch basin, which represents a segment. 1.3.10 MODEL BASED SEGMENTATION The central assumption of such an approach is that structures of interest/organs have a repetitive form of geometry. Therefore, one can seek for a probabilistic model towards explaining the variation of the shape of the organ and then when segmenting an image impose constraints using this model as prior. Such a task involves (i) registration of the training examples to a common pose, (ii) probabilistic representation of the variation of the registered samples, and (iii) statistical inference between the model and the image. State of the art methods in the literature for knowledge-based segmentation involve active shape and appearance models, active contours and deformable templates and level-set based methods.

1.3.11 MULTI-SCALE SEGMENTATION Image segmentations are computed at multiple scales in scale-space and sometimes propagated from coarse to fine scales; see scale-space segmentation. Segmentation criteria can be arbitrarily complex and may take into account global as well as local criteria. A common requirement is that each region must be connected in some sense. 1.3.12 ONE-DIMENSIONAL HIERARCHICAL SIGNAL SEGMENTATION Seminal work in scale space included the notion that a one-dimensional signal could be unambiguously segmented into regions, with one scale parameter controlling the scale of segmentation. A key observation is that the zero-crossings of the second derivatives (minima and maxima of the first derivative or slope) of multi-scale-smoothed versions of a signal form a nesting tree, which defines hierarchical relations between segments at different scales. Specifically, slope extrema at coarse scales can be traced back to corresponding features at fine scales. When a slope maximum and slope minimum annihilate each other at a larger scale, the three segments that they separated merge into one segment, thus defining the hierarchy of segments. 1.3.13 SEMI-AUTOMATIC SEGMENTATION In this kind of segmentation, the user outlines the region of interest with the mouse clicks and algorithms are applied so that the path that best fits the edge of the image is shown. Techniques like SIOX, Livewire, Intelligent Scissors or IT-SNAPS are used in this kind of segmentation

8 1.3.14 NEURAL NETWORKS SEGMENTATION Neural Network segmentation relies on processing small areas of an image using an artificial neural network or a set of neural networks. After such processing the decision-making mechanism marks the areas of an image accordingly to the category recognized by the neural network. A type of network designed especially for this is the Kohonen map. Pulse-coupled neural networks (PCNNs) are neural models proposed by modeling a cats visual cortex and developed for high-performance biomimetic image processing. PCNNs have been utilized for a variety of image processing applications, including: image segmentation, feature generation, face extraction, motion detection, region growing, noise reduction, and so on. A PCNN is a two-dimensional neural network. Each neuron in the network corresponds to one pixel in an input image, receiving its corresponding pixels color information (e.g. intensity) as an external stimulus. Each neuron also connects with its neighboring neurons, receiving local stimuli from them. The external and local stimuli are combined in an internal activation system, which accumulates the stimuli until it exceeds a dynamic threshold, resulting in a pulse output. Through iterative computation, PCNN neurons produce temporal series of pulse outputs of pulse outputs contain information of input images and can be utilized for various image processing applications, such as image segmentation and feature generation. PCNNs have several significant merits, including robustness against noise, independence of geometric variations in input patterns, capability of bridging minor intensity variations in input patterns, etc. 1.4 RETINAL BLOOD VESSEL SEGMENTATION The appearance and structure of blood vessels in retinal images play an important role in diagnosis of eye diseases. There includes various methods for segmentation of blood vessels in color retinal images. Diabetic eye disease refers to a group of eye problems that people with diabetes may face as a complication. Patients with diabetes are more likely to develop eye problems such as cataracts and glaucoma, but the disease's

9 affect on the retina is the main threat to vision. One of the complications of abnormalities in the retina and in the worst case blindness or severe vision loss is called Diabetic Retinopathy [DR].

Non-proliferative retinopathy is the less serious form of diabetic retinopathy and occurs when an abnormality develops in the retinal capillaries, allowing fluid to leak into the tissue of the eye. In this condition, a network of small blood vessels, called choroidal neovascularization (CNV), arises in the choroid and taking a portion of the blood supplying the retina. As the amount of blood supplying the retina is decreased, the sight may be degraded and in the severe cases, blindness may occur.

The most common signs of diabetic retinopathy include hemorrhages, cotton wool spots, dilated retinal veins, and hard exudates. Retinal images, also known as fundus or ocular images are acquired by making photographs of the back of the eye. Eye care specialists can screen large populations for vessel abnormalities after the development of an efficient and effective computer based approach to the automated segmentation of blood vessels in retinal images.

The detection and measurement of blood vessels can be used to classify the severity of disease, as part of the process of automated diagnosis of disease or in the assessment of the progression of therapy. Retinal blood vessels have measurable changes in diameter, branching angles, length, as a result of a disease. Thus a reliable method of blood vessel extraction and segmentation would be valuable for the early detection and characterization of changes due to such diseases.

10

CHAPTER 2 LITERATURE REVIEW

2.1 LINE OPERATORS AND SUPER VECTOR CLASSIFICATION

In reference [1] the framework of computer-aided diagnosis of eye diseases, retinal vessel segmentation based on line operators is proposed. A line detector, previously used in mammography, is applied to the green channel of the retinal image. It is based on the evaluation of the average grey level along lines of fixed length passing through the target pixel at different orientations. Two segmentation methods are considered. The first uses the basic line detector whose response is thresholded to obtain unsupervised pixel classification. As a further development, we employ two orthogonal line detectors along with the grey level of the target pixel to construct a feature vector for supervised classification using a support vector machine. The effectiveness of both methods is demonstrated through receiver operating characteristic analysis on two publicly available databases of color fundus images.

2.2 MATHEMATICAL MORPHOLOGY AND CURVATURE EVALUATION

This reference [2] presents an algorithm based on mathematical morphology and curvature evaluation for the detection of vessel-like patterns in a noisy environment. Such patterns are very common in medical images. Vessel detection is interesting for the computation of parameters related to blood flow. Its tree-like geometry makes it a usable feature for registration between images that can be of a different nature. In order to define vessel-like patterns, segmentation will be performed with respect to a precise model. We define a vessel as a bright pattern, piece-wise connected, and locally linear. Mathematical Morphology is very well adapted to this description, however other patterns fit such a morphological description. In order to differentiate vessels from analogous background patterns, a cross-curvature evaluation is performed. They are separated out as they have a

11 specific Gaussian-like profile whose curvature varies smoothly along the vessel. The detection algorithm that derives directly from this modeling is based on four Steps: 1) Noise reduction 2) linear pattern with Gaussian-like profile improvement 3) crosscurvature evaluation 4) linear filtering. We present its theoretical background and illustrate it on real images of various natures, then evaluate its robustness and its accuracy with respect to noise.

2.3 2D GOBAR WAVELET AND SUPERVISED CLASSIFICATION This reference [3] presents a method for automated segmentation of the vasculature in retinal images. The method produces segmentations by classifying each image pixel as vessel or nonvessel, based on the pixels feature vector. Feature vectors are composed of the pixels intensity and two-dimensional Gabor wavelet transform responses taken at multiple scales. The Gabor wavelet is capable of tuning to specific frequencies, thus allowing noise filtering and vessel enhancement in a single step. We use a Bayesian classifier with class-conditional probability density functions (likelihoods) described as Gaussian mixtures, yielding a fast classification, while being able to model complex decision surfaces. The probability distributions are estimated based on a training set of labeled pixels obtained from manual segmentations. The methods performance is evaluated on publicly available DRIVE and STARE databases of manually labeled images. On the DRIVE database, it achieves an area under the receiver operating characteristic curve of 0.9614, being slightly superior than that presented by state-of-theart approaches. We are making our implementation available as open source MATLAB scripts for researchers interested in implementation details, evaluation, or development of methods.

2.4 RIDGE BASED

A method is presented in reference [4] for automated segmentation of vessels in two-dimensional color images of the retina. This method can be used in computer analyses of retinal images, e.g., in automated screening for diabetic retinopathy. The

12 system is based on extraction of image ridges, which coincide approximately with vessel centerlines. The ridges are used to compose primitives in the form of line elements. With the line elements an image is partitioned into patches by assigning each image pixel to the closest line element. Every line element constitutes a local coordinate frame for its corresponding patch. For every pixel, feature vectors are computed that make use of properties of the patches and the line elements. The feature vectors are classified using a NN-classifier and sequential forward feature selection. The algorithm was tested on a database consisting of 40 manually labeled images. The method achieves an area under the receiver operating characteristic curve of 0.952. The method is compared with two recently published rule-based methods of Hoover and Jiang. The results show that our method is significantly better than the two rule-based methods ( <0.01). The accuracy of our method is 0.944 versus 0.947 for a second observer.

2.5 DIVERGENCE OF VECTOR FIELDS

In this reference [4], a method is proposed for detecting blood vessels in pathological retina images. In the proposed method, blood vessel-like objects are extracted using the Laplacian operator and noisy objects are pruned according to the centerlines, which are detected using the normalized gradient vector field. The method has been tested with all the pathological retina images in the publicly available STARE database. Experiment results show that the method can avoid detecting false vessels in pathological regions and can produce reliable results for healthy regions.

2.6 COMBINING THE DETECTION OF VESSEL CENTERLINES This reference [5] presents an automated method for the segmentation of the vascular network in retinal images. The algorithm starts with the extraction of vessel centerlines, which are used as guidelines for the subsequent vessel filling phase. For this purpose, the outputs of four directional differential operators are processed in order to select connected sets of candidate points to be further classified as centerline pixels using vessel derived features. The final segmentation is obtained using an iterative region

13 growing method that integrates the contents of several binary images resulting from vessel width dependent morphological filters. Our approach was tested on two publicly available databases and its results are compared with recently published methods. The results demonstrate that our algorithm outperforms other solutions and approximates the average accuracy of a human observer without a significant degradation of sensitivity and specificity. 2.7 TWO-DIMENSIONAL MATCHED FILTERS Although current literature in reference [6] abounds in a variety of edge detection algorithms, they do not always lead to acceptable results in extracting various features in an image. In this paper, we address the problem of detecting blood vessels in retinal images. Blood vessels usually have poor local contrast and the application of existing edge detection algorithms yield results which are not satisfactory. We introduce an operator for feature extraction based on the optical and spatial properties of objects to be recognized. The gray-level profile of the cross section of a blood vessel is approximated by a Gaussian shaped curve. The concept of matched filter detection of signals is used to detect piecewise linear segments of blood vessels in these images. We construct 12 different templates that are used to search for vessel segments along all possible directions. We discuss various issues related to the implementation of these matched filters. The results are compared to those obtained with other methods. The automatic detection of blood vessels in the retina could help physicians in diagnosing ocular diseases.

2.8 DETECTION AND TRACKING BY MATCHED GAUSSIAN AND KALMAN FILTERS The detection and tracking algorithms of the blood vessel network in the retinal images is proposed in reference [7]. Two main groups of algorithms are employed for this task, i.e., scanning and tracking. According to the known blood vessel feature, a second-order derivative Gaussian matched filter is designed and used to locate the center point and width of a vessel in its cross sectional profile. Together with this the Extended

14 Kalman Filter is employed for the optimal linear estimation of the next possible location of blood vessel segment by appropriate formulation of its pattern changing process and observation model. To check the bikcation in the vessel network, a simple branching detection strategy is implemented during tracking. The proposed algorithms all work well in the whole tracking process and can detect more complete vessel network in the ocular fundus photographs.

2.9 DETECTION AND QUANTIFICATION OF RETINOPATHY USING DIGITAL ANGIOGRAMS

An algorithm is presented for the analysis and quantification of the vascular structures of the human retina in reference [8]. Information about retinal blood vessel morphology is used in grading the severity and progression of a number of diseases. These disease processes are typically followed over relatively long time courses, and subjective analysis of the sequential images dictates the appropriate therapy for these patients. In this research, retinal fluorescein angiogram are acquired digitally in a 1024 x 1024 16-b image format and are processed using an automated vessel tracking program to identify and quantitate stenotic and/or tortuous vessel segments. The algorithm relies on a matched filtering approach coupled with a priori knowledge about retinal vessel properties to automatically detect the vessel boundaries, track the midline of the vessel, and extract useful parameters of clinical interest. By modeling the vessel profile using Gaussian functions, improved estimates of vessel diameters are obtained over previous algorithms. An adaptive densitometric tracking technique based on local neighborhood information is also used to improve computational performance in regions where the vessel is relatively straight.

2.10 AMPLITUDE MODIFIED SECOND ORDER GAUSSIAN FILTER

In this reference [9], the fitness of estimating vessel profiles with Gaussian function is evaluated and an amplitude-modified second-order Gaussian filter is proposed for the detection and measurement of vessels. Mathematical analysis is given and

15 supported by a simulation and experiments to demonstrate that the vessel width can be measured in linear relationship with the spreading factor of the matched filter when the magnitude coefficient of the filter is suitably assigned. The absolute value of vessel diameter can be determined simply by using a precalibrated line, which is typically required since images are always system dependent. The experiment shows that the inclusion of the width measurement in the detection process can improve the performance of matched filter and result in a significant increase in success rate of detection. 2.11 MEASUREMENT BASED ON 2-D MODELLING In the reference [10] Changes in retinal vessel diameter are an important sign of diseases such as hypertension, arteriosclerosis and diabetes mellitus. Obtaining precise measurements of vascular widths is a critical and demanding process in automated retinal image analysis as the typical vessel is only a few pixels wide. This paper presents an algorithm to measure the vessel diameter to subpixel accuracy. The diameter measurement is based on a two-dimensional difference of Gaussian model, which is optimized to fit a two-dimensional intensity vessel segment. The performance of the method is evaluated against Brinchmann-Hansens half height, Gregsons rectangular profile and Zhous Gaussian model. Results from 100 sample profiles show that the presented algorithm is over 30% more precise than the compared techniques and is accurate to a third of a pixel.

16

CHAPTER 3 EXISTING SYSTEM

Many methods for retinal vessel segmentation have been reported. These can be divided into two groups: rule-based methods and supervised methods. In the first group, we highlight methods using vessel tracking, mathematical morphology, matched filtering, model-based locally adaptive thresholding or deformable models. On the other hand, supervised methods are those based on pixel classification. 3.1 RULE-BASED METHODS 3.1.1 VESSEL TRACKING METHODS This method attempts to obtain the vasculature structure by following vessel center lines. Starting from an initial set of points established automatically or by manual labeling, vessels are traced by deciding from local information the most appropriate candidate pixel from those close to that currently under evaluation. 3.1.2 MATHEMATICAL MORPHOLOGY This method is to benefit from a priori-known vasculature shape features, such as being piecewise linear and connected. Then, by applying morphological operators, the vasculature is filtered from the background for final segmentation.

3.1.3 MATCHED FILTERING TECHNIQUE

This technique usually use a 2-D linear structural element with a Gaussian crossprofile section, extruded or rotated into three dimensions for blood vessel cross-profile identification (typically a Gaussian or Gaussian-derivative profile). The kernel is rotated into many different orientations (usually 8 or 12) to fit into vessels of different configuration. The image is then thresholded to extract the vessel silhouette from the background.

17

3.1.4 MODEL-BASED LOCALLY ADAPTIVE THRESHOLDING

This general framework based on a verification-based multithreshold probing scheme was presented by Jiang. These authors enriched this generic methodology by incorporating relevant information related to retinal vessels into the verification process with the aim of enabling its application to retinal images.

3.1.5 DEFORMABLE OR SNAKE MODELS A snake is an active contour model that, once placed on the image near the contour of interest, can evolve to fit the shape of the desired structure by an iterative adaption.

3.1.6 MULTISCALE FEATURE EXTRACTION

The local maxima over scales of the gradient magnitude and the maximum principal curvature of the Hessian tensor were used in a multiple pass region growing procedure. Growth progressively segmented the blood vessels by using both feature and spatial information.

3.1.7 LAPLACIAN AND GRADIENT VECTOR FIELD

In this method, blood vessel-like objects were extracted by using the Laplacian operator and noisy objects were pruned according to centerlines, detected by means of the normalized gradient vector field.

18 3.2 SUPERVISED METHODS

Supervised methods are based on pixel classification, which consists on classifying each pixel into two classes, vessel and non-vessel. Classifiers are trained by supervised learning with data from manually-labeled images.

3.2.1 BACK PROPAGATION MULTILAYER NEURAL NETWORK

This method is proposed for vascular tree segmentation. After histogram equalization, smoothing and edge detection, the image was divided into 2020 pixel squares (400 input neurons). The NN was then fed with the values of these pixel windows for classifying each pixel into vessel or not.

3.2.2 MULTILAYER PERCEPTRON NN

Each pixel in the image was classified by using the first principal component, and the edge strength values from a 1010 pixel subimage centered on the pixel under evaluation, as input data.

3.2.3 K-NEAREST NEIGHBOR (KNN) CLASSIFIER A 31-component pixel feature vector was constructed with the Gaussian and its derivatives up to order 2 at 5 different scales, augmented with the gray-level from the green channel of the original image.

3.2.4 SUPERVISED RIDGE- BASED VESSEL DETECTION METHOD

This assumps that vessels are elongated structures is the basis for the supervised ridge- based vessel detection method. Ridges were extracted from the image and used as primitives to form line elements. Each pixel was then assigned to its nearest line element, the image thus being partitioned into patches. For every pixel, 27 features were firstly computed and those obtaining the best class separability were finally selected.

19

3.2.5 KNN-CLASSIFIER SELECTION

AND

SEQUENTIAL

FORWARD

FEATURE

Feature vectors were classified by using a kNN-classifier and sequential forward feature selection. 3.2.6 GABOR WAVELET TRANSFORM Multiscale analysis was performed on the image by using this transform. The gray-level of the inverted green channel and the maximum Gabor transform response over angles at four different scales were considered as pixel features.

3.2.7 SUPPORT VECTOR MACHINE (SVM)

SVM is used for pixel classification as vessel or nonvessel. They used two orthogonal line detectors along with the gray-level of the target pixel to construct the feature vector.

20

CHAPTER 4 PROPOSED METHODOLOGY

This project presents precise measurement of retinal vessel diameter using two modules i.e., segmentation and width measurement. A new supervised method is proposed for the blood vessel segmentation in retinal images by using gray-level and moment invariants-based features. The change in width of retinal vessels within the fundus indicates the risk level of diabetic retinopathy. . After the segmentation of blood vessels of retina, width of the blood vessel is measured which inorder help to diagnose hypertension and cardiovascular diseases. This measurement of blood vessels in retinal image is done by a proposed method known to be graphic theoretic method.

To evaluate the vessel segmentation methodology described in the next section, two publicly available databases containing retinal images, the DRIVE and STARE databases, were used. These databases have been widely used by other researchers to test their vessel segmentation methodologies since, apart from being public; they provide manual segmentations for performance evaluation.

The STARE database is given as input for proposed retinal blood vessel segmentation, originally collected by Hoover, comprises 20 eye-fundus color images (ten of them contain pathology) captured with a Topcon TRV-50 fundus camera at 35 FOV. The images were digitalized to 700 605 pixels, 8 bits per color channel and are available in PPM format. The database contains two sets of manual segmentations made by two different observers. Performance is computed with the segmentations of the first observer as ground truth.

The database is divided into two sets: a test set and a training set, each of them containing 20 images. The test set provides the corresponding FOV masks for the images, which are circular (approximated diameter of 540 pixels) and two manual segmentations

21 generated by two different specialists for each image. The selection of the first observer is accepted as ground truth and used for algorithm performance evaluation in literature. The training set also includes the FOV masks for the images and a set of manual segmentations made by the first observer.

4.1 PROPOSED VESSEL SEGMENTATION METHOD

This proposes a new supervised approach for blood vessel detection based on a NN for pixel classification. The necessary feature vector is computed from preprocessed retinal images in the neighborhood of the pixel under consideration. The following process stages may be identified: 1) Original fundus image preprocessing for gray-level homogenization and blood vessel enhancement, 2) feature extraction for pixel numerical representation, 3) Application of a classifier to label the pixel as vessel or nonvessel, 4) Post processing for filling pixel gaps in detected blood vessels and removing falselydetected isolated vessel pixels.

4.2 BLOCK DIAGRAM FOR PROPOSED VESSEL SEGMENTATION METHOD

PRE PROCESSING

FEATURE EXTRACTION

CLASSIFICATION

INPUT IMAGE OUTPUT IMAGE

POST PROCESSING

Fig 1: Block diagram for vessel segmentation method

22

4.3 BLOCK DIAGRAM DESCRIPTION Input images are monochrome and obtained by extracting the green band from original RGB retinal images. The green channel provides the best vessel-background contrast of the RGB-representation, while the red channel is the brightest color channel and has low contrast, and the blue one offers poor dynamic range. Thus, blood containing elements in the retinal layer (such as vessels) are best represented and reach higher contrast in the green channel. The application of the methodology to retinas of different size (i.e., the diameter in pixels of STARE database retinas is approximately 650 pixels) demands either resizing input images to fulfill this condition or adapting proportionately the whole set of used parameters to this new retina size. Vessel segmentation undergo following steps with the input from STARE database from fundus images.

4.3.1 PREPROCESSING Color fundus images often show important lighting variations, poor contrast and noise. In order to reduce these imperfections and generate images more suitable for extracting the pixel features demanded in the classification step, a preprocessing comprising the following steps is applied: 1) vessel central light reflex removal. 2) Background homogenization. 3) Vessel enhancement. 4.3.1.1 VESSEL CENTRAL LIGHT REFLEX REMOVAL Since retinal blood vessels have lower reflectance when compared to other retinal surfaces, they appear darker than the background. Although the typical vessel crosssectional gray-level profile can be approximated by a Gaussian shaped curve (inner vessel pixels are darker than the outermost ones), some blood vessels include a light streak (known as a light reflex) which runs down the central length of the blood vessel. To remove this brighter strip, the green plane of the image is filtered by applying a morphological opening using a three-pixel diameter disc, defined in a square grid by

23 using eight-connexity, as structuring element. Disc diameter was fixed to the possible minimum value to reduce the risk of merging close vessels .I denotes the resultant image for future references.

Fig 2 shows the fragment of original image containing a vessel with central light reflex

Fig 3 shows the effect of light reflex removal

4.3.1.2 BACKGROUND HOMOGENIZATION Fundus images often contain background intensity variation due to nonuniform illumination. Consequently, background pixels may have different intensity for the same image and, although their gray-levels are usually higher than those of vessel pixels, the intensity values of some background pixels is comparable to that of brighter vessel pixels. Since the feature vector used to represent a pixel in the classification stage is formed by gray-scale values, this effect may worsen the performance of the vessel segmentation methodology. With the purpose of removing these background lightening variations, a shade-corrected image is accomplished from a background estimate. This image is the result of a filtering operation with a large arithmetic mean kernel.

24

Fig 4: Shows Homogenized images

1)33 mean filter is applied to smooth occasional salt-and-pepper noise. Noise smoothing is performed by convolving the resultant image with Gaussian kernel of dimensions mm=99, mean=0 and variance
2

=1.82,

2) A background image IB , is produced by applying a 6969 mean filter . When this filter is applied to the pixels in the FOV near the border, the results are strongly biased by the external dark region. To overcome this problem, out-of-the FOV graylevels are replaced by average gray-levels in the remaining pixels in the square. Then, the difference D between I and IB is calculated for every pixel. D(x,y) = I (x,y) - IB (x,y) (1)

3) A shade-corrected image ISC is obtained by transforming linearly RD values into integers covering the whole range of possible gray-levels (0 255), referred to 8-bit images).The proposed shade-correction algorithm is observed to reduce background intensity variations and enhance contrast in relation to the original green channel image. Besides the background intensity variations in images, intensities can reveal significant variations between images due to different illumination conditions in the acquisition process. In order to reduce this influence, a homogenized image IH is produced as follows: the histogram of ISC is displaced toward the middle of the gray-scale by modifying pixel intensities according to the following gray-level global transformation function:

25

Output=

(2)

where

Input+128-

Input_Max

(3)

Input

and

output

are the gray-level variables of input and output images (ISC and
Input_Max defines

IH respectively). The variable denoted by

the gray-level presenting the

highest number of pixels in ISC. By means of this operation, pixels with gray-level
Input_Max,

which are observed to correspond to the background of the retina, are set to

128 for 8-bit images. Thus, background pixels in images with different illumination conditions will standardize their intensity around this value.

4.3.1.3 VESSEL ENHANCEMENT The final preprocessing step consists on generating a new vessel-enhanced image (IVE), which proves more suitable for further extraction of moment invariants- based features. Vessel enhancement is performed by estimating the complementary image of the homogenized image IH, , and subsequently applying the morphological Top-Hat transformation. IVE= ( ) (4)

Where

is a morphological opening operation using a disc of eight pixels in radius.

Thus, while bright retinal structures are removed (i.e., optic disc, possible presence of exudates or reflection artifacts), the darker structures remaining after the opening operation become enhanced.

26

Fig 5: shows vessel-enhanced image

4.3.2 FEATURE EXTRACTION The aim of the feature extraction stage is pixel characterization by means of a feature vector, a pixel representation in terms of some quantifiable measurements which may be easily used in the classification stage to decide whether pixels belong to a real blood vessel or not. In this paper, the following sets of features were selected. Gray-level-based features: features based on the differences between the gray-level in the candidate pixel and a statistical value representative of its surroundings. Moment invariants-based features: features based on moment invariants for describing small image regions formed by the gray-scale values of a window centered on the represented pixels.

4.3.2.1 GRAY-LEVEL-BASED FEATURES Since blood vessels are always darker than their surroundings, features based on describing gray-level variation in the surroundings of candidate. pixels seem a good choice. A set of gray-level-based descriptors taking this information into account were derived from homogenized images IH considering only a small pixel region centered on the described pixel (x,y) . Sx,yw stands for the set of coordinates in a ww sized square window centered on point . Then, these descriptors can be expressed as

f1(x, y) = IH(x,y )-min {IH(s,t)}

(s,t)

(5) (6)

f2(x, y) = max {IH(s,t)}-IH(s,t)

(s,t)

27

f3(x, y) =IH(x, y)- mean {IH(s,t)}

(s,t)

(7)

f1(x,y)= std {IH(s,t)}

(s,t)

(8)

f5(x,y) =IH(x,y)
4.3.2.2 MOMENT INVARIANT BASED FEATURES

(9)

The vasculature in retinal images is known to be piecewise linear and can be approximated by many connected line segments. For detecting these quasi-linear shapes, which are not all equally wide and may be oriented at any angle, shape descriptors invariant to translation, rotation and scale change may play an important role. , moment invariants provide an attractive solution and are included in the feature vector. Given a pixel(x,y) of the vessel-enhanced image IVE , a subimage is generated by taking the region defined by . The size of this region was fixed to 17 17 so that,

considering that the region is centered on the middle of a wide vessel (8-9-pixel wide and referred to retinas of approximately 540 pixels in diameter), the subimage includes an approximately equal number of vessel and nonvessel pixels. For this subimage, denoted by .

A set of seven moment invariants under size, translation, and rotation, known as Hu moment invariants, can be derived from combinations of regular moments.
1= 20+ 02 2=

(10) +4
2 11

20+ 02)

(11)

the values describing vessel and non vessel central pixels become sensitive ,they reflect significant differences between both of them. Both
1

and

value increases comparing

the original one increases to describe if they are vessel pixels and decreases if they are

28 non vessel central pixel. In conclusion the descriptors were considered to be a part of the feature vector of a pixel located at (x,y) f6(x,y) =|log( f7(x,y) =|log(
1)| 2)|

(12) (13)

Fig: 6 Example of obtaining pixel environments for moment invariant calculation

To overcome the problem moments are computed on new subimage IHu produced by multiplying the original ones

p-1a |log( |log(

1)| 2)|

p-1b 4.73 11.29

p-2a 4.87 10.71

p-2b 5.02 11.81

p-3a 4.36 10.92

p-3b 4.23 10.90

p-4a 3.96 10.59

p-4b 3.92 12.11

5.26 11.70

Table 1: module of the

and

moments logarithm calculated from the subimages

p-1a |log( |log(

1)| 2)|

p-1b 2.89 9.16

1

p-2a 5.16 12.85

p-2b 3.13 11.11

p-3a 4.79 11.19

p-3b 2.34 8.31

p-4a 4.12 10.82

p-4b 2.21 7.79

5.34 13.57

Table 2: module of the

and

moments logarithm calculated from the subimages IHu

29

4.3.3 CLASSIFICATION In the feature extraction stage, each pixel from a fundus image is characterized by a vector in a 7-D feature space F(x,y)=(f1(x,y),..,f7(x,y)) (14)

Now, a classification procedure assigns one of the classes C1 (vessel) or C2 (nonvessel) to each candidate pixel when its representation is known. In order to select a suitable classifier, the distribution of the training set data (described below) in the feature space was analyzed. The results of this analysis showed that the class linear separability grade was not high enough for the accuracy level required for vasculature segmentation in retinal images. Therefore, the use of a non linear classifier was necessary. The following nonlinear classifiers can be found in the existing literature on this topic: the kNN method, support vector machines, Bayesian classifier, or neural networks . A multilayer feedforward NN was selected in this paper. Two classification stages can be distinguished: DESIGN STAGE, in which the NN configuration is decided and the NN is trained, APPLICATION STAGE, in which the trained NN is used to classify each pixel as vessel or nonvessel to obtain a vessel binary image.

4.3.3.1 NEURAL NETWORK DESIGN

A multilayer feedforward network, consisting of an input layer, three hidden layers and an output layer, is adopted in this paper. The input layer is composed by a number of neurons equal to the dimension of the feature vector (seven neurons). Regarding the hidden layers, several topologies with different numbers of neurons were tested. A number of three hidden layers, each containing 15 neurons, provided optimal NN configuration. The output layer contains a single neuron and is attached, as the remainder units, to a nonlinear logistic sigmoid activation function, so its output ranges between 0 and 1. This choice was grounded on the fact of interpreting NN output as

30 posterior probabilities The training set,ST , is composed of a set of candidates for which the feature vector [F (20)], and the classification result (C1 or C2 : vessel or nonvessel) are known

ST= {(F (n,),Ck(n))|n=1N; k

{1,2}}

(15)

The samples forming were collected from manually labeled nonvessel and vessel pixels in the DRIVE training images. Specifically, around 30 000 pixel samples, fairly divided into vessel and non-vessel pixels, were used. Unlike others selected their training set by random pixel-sample extraction from available manual segmentations of DRIVE and STARE images, we produced our own training set by hand. As in, gold-standard images may contain errors due to the considerable difficulty involved by the creation of these handmade images. To reduce the risk of introducing errors in and, therefore, of introducing noise in the NN, we opted for carefully selecting specific training samples covering all possible vessel, background, and noise patterns. It was applied to compute method performance with both DRIVE and STARE databases. Since the features of have very different ranges and values, each of these features is normalized to zero mean and unit variance independently by applying

=
where and



(16)

stand for the average and standard deviation of the feature calculated

over. Once is established, NN is trained by adjusting the weights of the connections through error interpretation. The back-propagation training algorithm was used with this purpose.

4.3.3.2 NEURAL NETWORK APPLICATION At this stage, the trained NN is applied to an unseen fundus image to generate a binary image in which blood vessels are identified from retinal background: pixels mathematical descriptions are individually passed through the NN. In our case, the NN input units receive the set of features provided by (5)(9) and (12) and (13), normalized according to (16). Since a logistic sigmoidal activation function was selected for the

31 single neuron of the output layer, the NN decision determines a classification value between 0 and 1. Thus, a vessel probability map indicating the probability for the pixel to be part of a vessel is produced. The bright pixels in this image indicate higher probability of being vessel pixel. In order to obtain a vessel binary segmentation, a thresholding scheme on the probability map is used to decide whether a particular pixel is part of a vessel or not. Therefore, the classification procedure assigns one of the classes (vessel) or (nonvessel) to each candidate pixel, depending on if its associated probability is greater than a threshold. Thus, a classification output image is obtained by associating classes and to the gray level values 255 and 0, respectively. (x,y) =  
(17)

where denotes the probability of a pixel described by feature vector to belong to class C1.

4.3.4 POST PROCESSING Classifier performance is enhanced by the inclusion of a two-step postprocessing stage: The first step is aimed at filling pixel gaps in detected blood vessels, The second step is aimed at removing falsely detected isolated vessel pixels.

Fig7: Postprocessing image

32 From visual inspection of the NN output, vessels may have a few gaps (i.e., pixels completely surrounded by vessel points, but not labeled as vessel pixels). To overcome this problem, an iterative filling operation is performed by considering that pixels with at least six neighbors classified as vessel points must also be vessel pixels. Besides, small isolated regions misclassified as blood vessel pixels are also observed. In order to remove these artifacts, the pixel area in each connected region is measured. In artifact removal, each region connected to an area below 25 is reclassified as nonvessel.

33

CHAPTER 5 EXPERIMENTAL RESULTS

5.1 PERFORMANCE MEASURES

In order to quantify the algorithmic performance of the proposed method on a fundus image, the resulting segmentation is compared to its corresponding gold-standard image. This image is obtained by manual creation of a vessel mask in which all vessel pixels are set to one and all nonvessel pixels are set to zero. Thus, automated vessel segmentation performance can be assessed. This algorithm was evaluated in terms of sensitivity (Se), specificity (Sp), positive predictive value (Ppv), negative predictive value (Npv) and accuracy (Acc). These metrics are defined as      =

(22)

Se and Sp metrics are the ratio of well-classified vessel and nonvessel pixels, respectively. Ppv is the ratio of pixels classified as vessel pixels that are correctly classified. Npv is the ratio of pixels classified as background pixels that are correctly classified. Acc is a global measure providing the ratio of total well-classified pixels. This algorithm performance was also measured with receiver operating characteristic (ROC) curves. A ROC curve is a plot of true positive fractions ( ) versus false positive fractions (1- ) by varying the threshold on the probability map. The closer a curve approaches the top left corner, the better the performance of the system. The area

34 under the curve (AUC), which is 1 for a perfect system, is a single measure to quantify this behavior.

5.2 PROPOSED METHOD EVALUTION

This method was evaluated on DRIVE and STARE database images with available gold-standard images. Since the images dark background outside the FOV is easily detected , , , and  values were computed for each image considering

FOV pixels only. Since FOV masks are not provided for STARE images, they were generated with an approximate diameter of 650 550. The average values of , ,


,

and 

are shown for the 20 images in each database. Threshold value

for all the images in the same database 0.91 for STARE images. The AUC measured

for ROC was 0.9769 for the STARE databases.

5.3 COMPARISION TO OTHER METHODS

In order to compare our approach to other retinal vessel segmentation algorithms, Acc and AUC were used as measures of method performance. Since these measurements were performed by other authors, this choice facilities comparing our results to theirs. The proposed method proves especially useful for vessel detection in STARE images. Its application to this database resulted in the second highest accuracy score among all experiments and the first when is the reference measurement. There are no available labeled training images for STARE images, Soares performed leave-one-out tests on this database. The STARE database contains ten images with pathologies, while the test of DRIVE only contains four. Moreover, abnormal regions are wider in STARE. Regarding performance comparison in terms of when results are jointly analyzed for DRIVE and STARE images, our algorithm renders greater accuracy than others authors algorithms, being outperformed only by Ricci and Perfettis proposal. This method proved very dependent on the training set, to research the dependence of their classification method on the dataset, carried out an experiment based on.

35 Firstly, training the classifier on each of the DRIVE and STARE databases, and then, testing it on the other. Their maximum accuracy values are can be observed that performance is worse now, since strongly decreases from 0.9595 to 0.9266 on DRIVE and 0.9646 to 0.9452 on STARE database images. Therefore, classifier retraining is necessary before applying their methodology on a new database. Training the classifier with STARE images, the resulting values are shown to facilitate comparisons between both methods under identical conditions. In this case, it is clearly observed that our estimated performance in terms of method accuracy is higher, thus proving higher training set robustness.

FIG 8: INPUT IMAGE

FIG 9: PROPOSED OUTPUT IMAGE

36

CHAPTER 6 CONCLUSION

This method is based on a NN scheme for pixel classification, being the feature vector representing each pixel composed of gray-level and moment invariants-based features. The experiments aimed at evaluating the efficiency of the applied descriptors prove this method is capable of rendering accurate results, even when these types of features are used independently. Thus, accuracy improves up 0.9526 for the 20 test images in the STARE databases. Therefore, the method finally adopts a 7-D feature vector composed by the five gray-level and the two moment invariants-based features. The proposed method uses a NN for pixel classification as vessel or non-vessel. This classifier was selected after method accuracy assessment by means of a kNN and a SVM, instead of a NN. This NN showed better accuracy than a kNN and a SVM for all cases. Its application to the STARE database results in the second highest accuracy score among all experiments. In addition, method simplicity should also be highlighted. Its pixel classification procedure is based on computing only seven features for each pixel, thus needing shorter computational time. The total time required to process a single image is less than approximately one minute and thirty seconds, running on a PC with an Intel Core2Duo CPU at 2.13 GHz and 2 GB of RAM. Since our implementation is experimental, this performance might still be improved. The demonstrated effectiveness and robustness, together with its simplicity and fast implementation, make this proposed automated blood vessel segmentation method a suitable tool for being integrated into a complete prescreening system for early DR detection. After segmentation of retinal blood vessel, the segmented output is proposed to measure the width of the retinal blood vessel. This second module will be continued in the next phase. The change in retinal blood vessel diameter indicates that the vessel is affected by DR.

37 CHAPTER 7 REFERENCES

[1] E. Ricci and R. Perfetti, Retinal blood vessel segmentation using line operators and support vector classification, IEEE Trans. Med. Imag., vol. 26, no. 10, pp. 13571365, Oct. 2007.

[2] F. Zana and J. C. Klein, Segmentation of vessel-like patterns using mathematical morphology and curvature evaluation, IEEE Trans. Image Process., vol. 10, no. 7, pp. 10101019, Jul. 2001.

[3] J. V. B. Soares, J. J. G. Leandro, R. M. Cesar, Jr., H. F. Jelinek, and M.J. Cree, Retinal vessel segmentation using the 2D Gabor wavelet and supervised classification, IEEE Trans. Med. Imag., vol. 25, no. 9, pp. 12141222, Sep. 2006.

[4]J. Staal, M. D. Abrmoff, M. Niemeijer, M. A. Viergever, and B. v. Ginneken, Ridge based vessel segmentation in color images of the retina, IEEE Trans. Med. Imag., vol. 23, no. 4, pp. 501509, Apr. 2004.

[5] B. S. Y. Lam and H. Yan, A novel vessel segmentation algorithm for pathological retina images based on the divergence of vector fields, IEEE Trans. Med. Imag., vol. 27, no. 2, pp. 237246, Feb. 2008.

[6] A. M. Mendona and A. Campilho, Segmentation of retinal blood vessels by combining the detection of centerlines and morphological reconstruction, IEEE Trans. Med. Imag., vol. 25, no. 9, pp. 12001213, Sep. 2006.

38 [7] S. Chaudhuri, S. Chatterjee, N. Katz, M. Nelson, and M. Goldbaum, Detection of blood vessels in retinal images using two-dimensional matched filters, IEEE Trans. Med. Imag., vol. 8, no. 3, pp. 263269, Sep. 1989.

[8] O. Chutatape, L. Zheng, and S. Krishman, Retinal blood vessel detection and tracking by matched Gaussian and Kalman filters, in Proc. IEEE Int. Conf. Eng. Biol. Soc., 1998, vol. 20, pp. 31443149.

[9] L. Zhou, M. S. Rzeszotarski, L. J. Singerman, and J. M. Chokreff, The detection and quantification of retinopathy using digital angiograms, IEEE Trans. Med. Imag., vol. 13, no. 4, pp. 619626, Dec. 1994. [10] L. Gang, O. Chutatape, and S. M. Krishnan, Detection and measurement of retinal vessels in fundus images using amplitude modified second-order Gaussian filter, IEEE Trans. Biomed. Eng., vol. 49, pp. 168172, Feb. 2002.

[11] J. Lowell, A. Hunter, D. Steel, A. Basu, R. Ryder, and R. L. Kennedy, Measurement of retinal vessel widths from fundus images based on 2-D modeling, IEEE Trans. Med. Imag., vol. 23, no. 10, pp. 11961204, Oct. 2004.