Review on Different Segmentation Techniques For Lung Cancer CT Images Arathi 1, Anusha Shetty 1, Madhushree 1, Chandini Udyavar 1, Akhilraj.V.Gadagkar 2 1 UG student, Dept. Of CSE, Srinivas school of engineering, Mangalore, 2 Asst.Prof.Dept. Of CSE, Srinivas school of engineering, Mangalore, Abstract Lung cancer is a disease characterized by uncontrolled growth of cell in tissues of the lung. Most cancers that start in the lung, known as primary lung cancers, are carcinomas. The two main type are Small cell lung carcinoma(sclc) and Non Small cell lung carcinoma(nsclc). Lung Computer-Aided Diagnosis (CAD) is a potential method to accomplish a range of quantitative tasks. For identifying the lung diseases, computed tomography (CT) scan is widely applied in diagnose. In this paper we proposed lung segmentation technique to accurately segment the lung parenchyma of lung CT images, which can help radiologist in early diagnosing lung diseases. Thus an organized review on image segmentation methods is essential and this paper provides a review on the various image segmentation techniques proposed in the literature. Keywords Lung cancer, Computed tomography, Lung carcinoma, Image segmentation. I. INTRODUCTION Lung cancer is the uncontrolled growth of abnormal cells that start off in one or both lungs.. The incidence of lung cancer is strongly correlated with cigarette smoking with result of tobacco use and also by air pollution. The most common symptoms of lung cancer are :A cough, chest pain,weight loss, infections such as bronchitis and pneumonia. Some of the existing diagnosis methods used in lung cancer are: X-ray, CT scan, MRI, Ultrasound, Endoscopy, PET scan. Most lung cancer have already spread widely and are at an advanced stage when they are first found. These cancer are very hard to cure. But in recent years, doctors have found tests that can be used to screen for lung cancer in people at high risk of the disease. This helps to find some of these cancer early, which can lower the risk of dying from this diseases. Image processing is used to detect the lung cancer.image segmentation is very important and challenging processes of image processing. The main goal of an image segmentation is to divide an image into several parts/segments having similar features or attributes. The image segmentation technique used in this way : Thresholding, deformable,region based and so on. II. LUNG CANCER SEGMENTATION AND PREDICTION TECHNIQUESR The segmentation accuracy directly affects many aspects, such as the malignancy classification of lung nodules in CAD for feature extraction. In section of paper we study various segmentation techniques for lung nodules from images. 2.1 Thresholding Thresholding methods are the simplest methods for image segmentation. These methods divide the image pixels with respect to their intensity level. These methods are used over images having lighter objects than background. The selection of these methods can be manual or automatic i.e. can DOI:10.23883/IJRTER.2018.4038.8JR8F 323
be based on prior knowledge or information of image features. Thresholding operation first converts the grey scale image into binary image. A threshold value T is selected in thresholding operation and it assigns two levels to the images that is one is above and the other is below the threshold value. By using the threshold value T, we can separate the object from the background. Then any point (x, y) for which f(x, y) > T is called an object point, otherwise the point is called a background point. There are basically three types of thresholding. a)global Thresholding: This is done by using any appropriate threshold value/t. This value of T will be constant for whole image. On the basis of T the output image can be obtained from original image as:. b) Variable Thresholding: In this type of thresholding, the value of T can vary over the image. This can further be of two types 1. Local Threshold: In this the value of T depends upon the neighborhood of x and y. 2. Adaptive Threshold: The value of T is a function of x and y. c) Multiple Thresholding : In this type of thresholding, there are multiple threshold values like T0 and T1. By using these output image can be computed as: The values of thresholds can be computed with the help of the peaks of the image histograms. Advanatges of Thresholding It is simplest method for image processing. It is a most significant tool for image processing. @IJRTER-2017, All Rights Reserved 324
The values of thresholds can be computed with the help of the peaks of the image segmentation. Disadvantages of Thresholding No guarantee of object coherency-may have holes, extraneous pixels, etc Highly dependent are not peaks, spatial details are not considered. 2.2. Mathematical Morphology (MM) To fill in holes and small gaps in the image morphological closing operation is applied on the threshold image. It first reserve the block whose area is the largest and then set the others to zero using 8-connected neighbors. Using the above step binary lung mask is obtained. To Extract the lung edge set a pixel to 0 if its 4-connected neighbors are all 1 s, this leaving only edge pixels. Original Lung CT image is multiplied with the lung masked image to get the final segmented lung region with gray level values as those of original image. Advantages of Mathematical Morphology: Mathematical morphology was also used for detection lung nodules in 3D CT images. It is a better method used for edge information detecting and noise filtering then differential equation. It is a better method used between noise smoothing. Disadvantages of Mathematical Morphology: It is not used on traditional mathematical model and analysis. Computation is more complex then general morphological edge detection. 2.3. Region Based Segmentation The region based segmentation methods are the methods that segments the image into various regions having similar characteristics. There are three basic techniques based on this method. I. Region Based Segmentation Region-based technique partitions the image into regions. This method works on the principle of homogeneity by considering the fact that the neighboring pixels inside a region possess similar characteristics and are dissimilar to the pixels in other regions. This technique divides an image into different regions based on the pre-defined criteria include color, intensity, or object. A region is, as opposed to an edge, a global concept and is formed by a closed path. The objective of region based segmentation is to produce a homogeneous region which is bigger in size and a provision to note any considerable changes in the characteristic of the neighboring pixels. The various types of region based segmentation include region growing, region splitting and merging and graph based methods. II. Region growing The basic idea of this method is to group a collection of pixels in an image with similar properties to form a region. In this method region grows by choosing a starting point called seed pixel. Then, the region grows by adding similar neighboring pixels according to a certain homogeneity criterion, increasing step by step the size of the region. Region growing can be processed in four steps: (i) Mark the group of seed pixels in original image. (ii) Select a clustering criterion such as grey level intensity or color and set up a stopping rule. @IJRTER-2017, All Rights Reserved 325
(iii) Expand the regions by connecting to each seed to the neighboring pixels that have satisfied the cluster properties similar to seed pixels. (iv) Stop region growing when no more pixels meet the criterion for inclusion in that region. III. Region splitting and merging In region splitting and merging technique an image is subdivided into a set of arbitrary unconnected regions. This method attempts to divide an input image into number of smaller regions recursively. This method initially considers the entire image as one single region and then divides the image into four quadrants based on certain pre-defined criteria. The advantage of this technique is to make the complete use of pixels relationship based on the image properties. The disadvantage of this technique is the selection of pixels based within the region since an over stringent criterion creates fragmented regions and a lenient criterion overlook blurred regions. Let "p" be the original image and T be the particular predicate. First of all the R1 is equal to p. Each region is divided into quadrants for which T (Ri) = False. If for every region, T (Rj) = True, then merge adjacent regions Ri and Rj such that T (Ri U Rj) = True Repeat step 3 until merging is impossible. 1 st split 2 nd split merge Whole image Fig: division of regions based on quad tree Advantages of Region based : More immune to noise, useful when it is easy to define similarity criteria. Connected region are guaranteed and IQM reduces lengthy neighbor problems during merging. Disadvantages of Region based : Expensive method in term of time and memory. The position and orientation of the image leads to blocky final segmentation and Regular division leads to over segmentation by splitting. The drawback can be overcome by using normalized cut @IJRTER-2017, All Rights Reserved 326
2.4 Deformable Model Deformable models are curves or surfaces defined within an image domain. These can move under the influence of internal forces, which are defined within the curve or surface, and external forces, which are computed from the image data. The internal forces are intended to keep the model smooth during deformation. The external forces are intended to move the model toward an object edge or other needed features within an image.such a edge description can then be readily used by subsequent applications. Advantages Of Deformable : It is widely applied method for 3D segmentation purpose It is a multiphase level set framework. Active contours approach. Disadvantages Of Deformable : It is complex method because it use lots of mathematical operation. When deformable registration is used to map dose between scans registration errors. Yield error in the dose associated with the mapped location. 2.5 Watershed Based s The watershed based methods uses the concept of topological interpretation. In this the intensity represents the basins having hole in its minima from where the water spills. When water reaches the border of basin the adjacent basins are merged together. To maintain separation between basins dams are required and are the borders of region of segmentation. These dams are constructed using dilation. The watershed methods consider the gradient of image as topographic surface. The pixels having more gradient are represented as boundaries which are continuous. Advantages Of Watershed : Results are more stable, detected boundaries are continuous. The union of all the regions forms the entire regions. The resulting boundaries form closed and connected regions. Disadvantages Of Watershed : Complex calculation of gradient. Watershed is most a natural images it produces excessive over segmentation. 2.6 Edge Based Segmentation The edge detection techniques are well developed techniques of image processing on their own. The edge based segmentation methods are based on the rapid change of intensity value in an image because a single intensity value does not provide good information about edges. In edge based segmentation methods, first of all the edges are detected and then are connected together to form the object boundaries to segment the required regions. The basic two edge based segmentation methods are: Gray histograms and Gradient based methods. Result of these methods is basically a binary image. These are the structural techniques based on discontinuity detection. @IJRTER-2017, All Rights Reserved 327
Advantages Of Edge Based : Good for images having better contrast between objects. Edge detection is one of the structural techniques of the image segmentation. Some of the edges are easy to find. Disadvantages Of Edge Based : Not suitable for wrong detect or too many edges. There is no any common performance to judge the edge detection techniques. Table1.COMPARISION OF VARIOUS SEGMENTATION TECHNIQUES Segmentation technique Thresholding method Edge Region Based Based Watershed Morphological Deformable Description Advantages Disadvantages Based on the histogram peaks of the image to find particular threshold values Basedon discontinuity detection Based on partitioning image in to homogeneous regions. Based on topological interpretation Based on threshold image. Based on medical image segmentation. No need of previous information,simplest method Good for images having better contrast between objects More immune to noise,useful when it is easy to define similarity criteria. Results are more stable, detected boundaries are continuous. It is better method for edge information detecting and noise filtering the differential operation.it is better method between noise smoothing. It is widely applied method for 3D segmentation purpose. Multiphase level set framework. Active contours approach Highly dependent o are not peaks, spatial details are not considered. Not suitable for wrong detected or too many edges. Expensive method interms of time and memory. Complex calculation of gradient. It is not used on traditional mathematical model and analaysis. Computation is more complex than general morphological edge detection. It is a complex method because it uses lots of mathematical operations. @IJRTER-2017, All Rights Reserved 328
III.CONCLUSION This paper Summarizes various segmentation techniques. Segmentation can be applied to any type of image. The deformable method is introduced to overcome short comings of region based methods. Since the data set consist of multiple images a method for initializing active contour in consecutive images is introduced. Result obtained by deformable method are good and method offers significant help to radiologists who need to analyze a set of CT images but still this method is complex because it uses lots of mathematical operation. Comparing to other methods Thresholding is simplest and computationally fast. Depending on the application technique varies. REFERNCES I. Sluimer, A. Schilham, M. Prokop, and B.Van Ginneken (2006), Computer analysis of computed tomography scans of the lung: a survey, IEEE Transactions on Medical Imaging, vol. 25, no. 4, pp. 385 405. II. en.wikipedia.org (2014), Region Based Segmentation, Available at :http://en.wikipedia.org/wiki/region growing. III. K. Castleman, "Digital Image Processing", Prentice Hall, 1996. IV. P. Dhawan, "Medical Image Analysis", IEEE press series in Biomedical Engineering, John Wiley & Sons, 2003. V. J. Roerdink, A. Meijster, "The Watershed Transform: Definitions, Algorithms and Parallelization Strategies", Fundamenta Informaticae, pp. 187-228, 2001. VI. VII. VIII. IX. Gajdhane, M. V. A., & Deshpande, L. M. Detection of Lung Cancer Stages on CT scan Images by Using Various Image Processing Techniques. Gomathi, M., & Thangaraj, P. (2010). A computer aided diagnosis system for lung cancer detection using support vector machine. American Journal of Applied Sciences, 7(12), 1532. Ilya Levner, Hong Zhangm(2007), Classification driven Watershed segmentation, IEEE TRANSACTIONS ON IMAGE PROCESSING VOL. 16, NO. 5. Sunil Kumar(2014), Lung Segmentation using Region Growing Algorithm, International Journal of Advanced Research in Computer Science and Software Engineering Volume 4. @IJRTER-2017, All Rights Reserved 329