04-09-2014, 10:02 AM
2Motivation
Motivation.docx (Size: 754.98 KB / Downloads: 8)
1 INTRODUCTION
Identifying specific organs or other features in medical images requires a considerable amount of expertise concerning the shapes and locations of anatomical features. Such segmentation is typically performed manually by expert physicians as part of treatment planning and diagnosis. Due to the increasing amount of available data and the complexity of features of interest, it is becoming essential to develop automated segmentation methods to assist and speed-up image-understanding tasks. Medical imaging is performed in various modalities, such as magnetic resonance imaging (MRI),computed tomography (CT), ultrasound, etc. Several automated methods have been developed to process the acquired images and identify features of interest, including intensity-based methods, region-growing methods and deformable contour models . Intensity-based methods identify local features such as edges and texture in order to extract regions of interest. Region-growing methods start from a seed-point (usually placed manually) on the image and perform the segmentation task by clustering neighborhood pixels using a similarity criterion. Deformable contour models are shape-based feature search procedures in which a closed contour deforms until a balance is reached between its internal energy (smoothness of the curve) and external energy (local region statistics such as first and second order moments of pixel intensity). Such methods are typically based on only one image feature, such as texture, shape, pixel intensity, etc. However, due to the low contrast information in medical images, an effective segmentation often requires extraction of a combination of features such as shape and texture or pixel intensity and shape. This paper describes our attempts to develop a segmentation algorithm that incorporates both shape and textural information to delineate a desired object in an image. In particular, motivated by the work of Harvey et al. and Tsai et al. we developed a genetic algorithm for medical image segmentation. The genetic algorithm framework brings considerable flexibility into the segmentation procedure by \incorporating both shape and texture information. In the following sections we describe our algorithm in depth and relate our methodology to previous work in this area. We start by reviewing the active shape modeling approach for image segmentation,specifically the level set method of shape characterization. We then describe texture-based segmentation methods such as Laws’ textural feature extraction method. We also describe our studies of the GENIE system for multi-spectral feature extraction. Finally, we present the results of our method on segmenting the prostate based on a small training set of pelvic CT images. We compare these results with those from similar runs on GENIE.
1.2Motivation
Image segmentation is an important process and its results are used in many image processing applications. However, despite its importance, there doesn’t to seem to beany general method of image segmentation that works well on all images. Imagesegmentation involves a lot of uncertainty, often with many parameters that need tobe tuned to provide optimal results. For example, the Phoenix image segmentationmethod has 14 adjustable parameters. This large number of parameters creates avery large search space. Color images have even more information than grey-scale images, and this information can be used to create higher quality segmentation. Itdoes, however, increase the complexity of the problem. A way of handling the large search space is to use a directed search method, such as genetic algorithms. Genetic algorithms have many qualities that make them well suited to the problemof image segmentation, such as the ability to forego a local optimum to reach a globaloptimum and the ability to efficiently find an optimal solution from within a large search space.Genetic algorithms could allow an image segmentation process that usually requiresmanual input to become unsupervised. Genetic algorithms have been used to successfullycolor segment images. Dueto their flexibility, it seems feasible to be able to use them to come up with a generalsegmentation method.
2Image Segmentation
Image segmentation is the process of dividing an image into homogeneous regions. This is equivalent to finding the boundaries between the regions. Segmentation isthe first step for many higher level image processing and computer vision opera-tions, including shape recognition, medical imaging[24], locating objects in satellite images[33], face detection[4] and road sign recognition[48].
2.3Methods of Image Segmentation
Image segmentation is an old and important problem, and there are numerous imagesegmentation methods. Most of these methods were developed to be used on ascertain class of images and therefore aren’t general image segmentation methods.Bhanu and Lee divide the image segmentation algorithms into three major
2.4Categories:
1. Edge Based
2. Region Based
3. Clustering Based
4. Model Based
1. Edge Based Techniques
Edge detection involves the detection of boundaries between different regions ofthe image. These boundaries correspond to discontinuities between pixels of thechosen feature (e.g.color, texture, intensity).
2.Region Based Techniques
Region splitting is an image segmentation method whereby pixels are classifiedinto regions. Each region corresponds to a range of feature values, with thresholdsbeing the delimiters. The choice of these thresholds is very important, as it greatlyaffects the quality of the segmentation. This method tends to excessively split regions, resulting in over segmentation .Region growing joins neighboring pixels with similar characteristics to form larger regions. This continues until the termination conditions are met. Most of theregion growing algorithms focus on local information, making it difficult to get goodglobal results. This method tends to excessively add to regions, resulting in under segmentation .Region merging recursively merges similar regions. It is similar to region grow-ing, except that two whole regions are combined, rather than one region combiningwith individual pixels. Region splitting and merging tries to overcome the
Parameter Modification
Most image segmentation methods have many parameters, constants and thresholds that need to be adjusted to produce optimal segmentation results. This creates avery large search space. Since the parameters typically interact in complex and non-linear ways, an analytic solution is not generally possible. With a reasonable amountof computation, genetic algorithms are able to find good approximations of a globaloptimum within a large search space. They are therefore well suited to problems involving parameter optimization. Most of the applications of genetic algorithms toimage segmentation involve the optimization of various parameters. Bhanget appose image segmentation as an optimization problem. They de-fine a general segmentation method, whereby genetic algorithms are applied to theparameters of various well known image segmentation methods. They advocate theuse of genetic algorithms to adapting the parameters of knows segmentation methods in order to be applicable to general images. They used outdoor color imagery and adapted 4 parameters of the Phoenix segmentation algorithm with genetic algorithms. They had successful results, producing high quality image segmentationwith a reasonable amount of computation. Even though they perform well on out-door scenes, these algorithms have not been proved to be able to cope with generalimages. The fact that these algorithms can be modified to adapt the parameters ofother segmentation methods makes this method very promising.Feitosa et al adopt a very similar approach and use genetic algorithms to modifythe parameters of a region merging segmentation algorithm. They use a fitnessfunction that measures the similarity of resulting segments to a target segmentationprovided by a user. Though computation is straight forward and intuitive, manualsegmentation is still necessary beforehand. This method can easily be adapted to
modify parameters of other segmentation methods.Zingaretti el al propose using genetic algorithms in unsupervised color imagesegmentation. This is another case of parameters of an existing image segmentationmethod being tuned by genetic algorithms. A key difference in this method isthat it performs multi-pass thresholding. Different thresholds are adapted during each pass of genetic algorithms. An important advantage of this method over theprevious one is that segmentation is performed totally unsupervised, without anymanual segmentation. It also doesn’t rely on any prior information regarding the type of image that is being processed or the task for which the segmentation resultswill be used. This approach successfully segmented a wide variety of images, with the exception of images that were highly textured.Pignalberi et al[39] use genetic algorithms for the optimization of parameters in animage segmentation algorithm. In this case, they focused on range images, where apixel is coloured depending on the distance between the object and a sensor. Thismethod segments out surfaces of 3D objects, but could be applied to segmentationof 2D images.
Modified Genetic Algorithms
Gong and Yang represent the image and the segmentation results by quadtrees.In a similar way to Zingaretti el al, they define a two pass system, genetic algorithms being used for optimization in both passes. In the first pass, genetic algorithms are used to minimize an energy function. In the optional second pass, a parameter defining how coarse or fine the segmentation is modified by geneticalgorithms to obtain optimal segmentation results. The chromosomes encode thequadtrees, making it inefficient to apply the usual crossover and mutationoperations. To cope with this, a new crossover method and three mutation methods are defined.\Aoyagi and Tsuji[7] use modified genetic algorithms for pixel-level segmentation. They approach image segmentation as a feature clustering problem and like Gong andYang use an energy function as a fitness function. They found it difficult toget ideal segmentation using traditional genetic algorithms, and so introduced forspecial types of mutation. They also propose a new method for creating individuals\of the population.
CONCLUSIONS AND FUTURE WORK
The algorithm developed here evolves a segmenting contour by incorporating both texture and shape information to extract objects without prominent edges, such as the prostate on pelvic CT images. Representing the shape of the contours as level sets and encoding candidate solutions of the GA as segmenting contours eliminates the need for deriving the gradients of energy functions for shape evolution and simplifies the optimization procedure. Our experiments using a small training set and a small population of candidate segmentation contours shows promise by converging on the prostate area. The following enhancements to the above framework are proposed for improving the segmentation results.
1. Incorporating position information: The relative position of the various organs, if incorporated, can be used for initial placement of the segmenting curve (which israndom at present). This has the potential to significantly improve the segmentation results
2. Extension to 3-D: Following the lead of [20], the pose parameters can be extended to represent the 3D pose of an object. The above framework can be used to evolve a surface instead of a curve in a 3-D domain. Thus information from all the slices of a CT scan can be used simultaneously for 3-D segmentation. We would also like to compare our results with the shape-based segmentation procedure implemented by Tsai et al