You are on page 1of 5
 
LUNG CANCER AND LUNG NODULES- AN INTRODUCTION
1
Afshan Khanum,
2
Purushothaman S. and
3
Rajeswari P.
1
Research Scholar, Department of CSE, VELS University, Chennai, India. Email:meena291198@gmail.com
2
Associate Professor, Email:dr.s.purushothaman@gmail.com
3
Lecturer, Department of Electrical and Computer Science Engineering, Email:rajeswaripuru@gmail.com
23
Institute of Technology, Haramaya University, DireDawa, Ethiopia
Abstract:
This paper presents an introduction to lung cancer and lung nodules. Some of the basic segmentation techniques have been presented.
1. LUNG CANCER
Lung nodules in human bodies indicate the lung abnormalities. A pulmonary nodule is the most common manifestation of lung cancer. Lung nodules are approximately spherical regions of relatively high density that are visible in X-ray images of the lung. Large malignant nodules with diameter greater than 1 centimeter can be detected using the traditional imaging equipment easily. Such large nodules can be diagnosed by needle biopsy or  bronchoscopy techniques. However, only few diagnostic options are available for the effective detection of small malignant nodules due to complexity associated in accessing small tumors, if they are located inside the tissue. Hence, additional diagnostic and imaging techniques are needed to solve this problem. One of the most promising techniques for detecting small cancerous nodules relies on characterizing the nodule, based on its growth rate. The growth rate is estimated by measuring the volumetric change of the detected lung nodules over time. Therefore, it is important to accurately measure the volume of the nodules to quantify their growth rate over time. Early detection of nodules can help in saving lung patients. Lung nodules can be detected by radiologists based on their experiences in examining the lung images. Nodules and nodular patterns are seen both in chest radiographs and in CT scans.
Figure 1.1 Anatomy of Human Lungs
A single nodule has the appearance of a rounded or irregular opacity, which may be defined as solid, non-solid or partly solid; and soft tissue or Ground-Glass Opacity (GGO) which is less than 3 cm diameter. Due to all these factors, automatic lung nodule detection from the CT images is still a challenging issue. Moreover, a number of mathematical algorithms have been proposed in the past. They must be supported by a good radiologist in order to make effective decisions. Chest image processing is a good approach for effective pulmonary image analysis. The CT images have  been used by many radiologists for the diagnosis of various pulmonary diseases such as lung cancer, Tuberculosis and Pulmonary embolism.
International Journal of Applied Engineering Research, ISSN 0973-4562 Vol.9 No.27 (2014)© Research India Publications; httpwww.ripublication.comijaer.htm9611
 
The lungs are a pair of involuntary organs present in the chest which performs multitude of various important functions in each and every second of our lives. Breathing is the most essential operation for these functions in which the lungs takes the oxygen from the atmosphere, transports it into the bloodstream and removes the carbon dioxide from the bloodstream towards the atmosphere. Human lungs are branched into right and left lungs. The left lung is divided into two lobes and the right into three lobes. The tissues present in the lung are spongy in nature due to the presence of gas filled cavities called alveoli. The anatomy of human lungs is shown in the Figure 1.1. There are about 250 -350 million alveoli in an adult lung. The lungs play a vital role in the body's defensive mechanism against the infection and other harmful environmental factors Lung cancer is caused by the rapid growth and division of cells which takes place in the formation of lungs. This is one of the most lethal and common form of cancers worldwide. In the past, researchers showed that this type of cancer affects men. However, in the last few decades, the lung cancer rate for women have been found to be increasing due to the rise in number of female smokers. The lung Cancer can be classified into two major categories namely primary lung cancer and secondary (metastatic) lung cancer. The primary lung cancer originates in the cells of the lungs, while secondary spread to lungs from other parts of the body commonly from the breast, colon, prostate, kidney, thyroid gland, stomach, cervix, rectum, testis, bone and skin (melanoma). There are two main types of lung carcinoma, namely Small Cell Lung Carcinoma (SCLC) and Non-Small Cell Lung Carcinoma (NSCLC). These lung carcinomas are categorized  by the size and appearance of the malignant cells under the microscope. The NSCLC are further classified into squamous cell carcinoma, large cell carcinoma and adenocarcinoma. The SCLC (20% of all cases) is less common, more aggressive and faster than NSCLC (80% of all cases) [https://s3-us-west-1.amazonaws.com/static.cancercommons.org/Handbook_v4 05_electronic+copy.pdf ] A tissue with abnormal growth of cells is called as a tumor. The tumor was classified in to two major categories namely Malignant (cancerous) and benign (non-cancerous). The main differences between these two types of tumors are the benign tumor that grows slower and this is rarely life threatening and will not spread to other organs present in the human body. These types of tumors will not come back usually when it is removed surgically. Malignant tumors are life threatening and grows back even after removing it surgically. It also spreads quickly to other organs present in the body. The methods used for effective diagnosis of lung cancer include Computed Tomography scan, Magnetic Resonance Imaging (MRI) scan, Positron Emission Tomography (PET) scan,  bronchoscopy (examination of the airways with fiber optics) and biopsy (examination of lung tissue sample). However, the biopsy is used to detect the type of cancer depending on their appearance under the microscope.
1.2 MEDICAL IMAGE SEGMENTATION
Segmentation is the process of highlighting the boundaries of objects of interests in a given image based on their shapes, locations of the objects, and the depth of each pixel. Medical image segmentation helps in differentiating good tissues of an organ from the other parts of a body and from a damaged structure like bone  breakage, or growth of tumor in brain, breast etc.
 
Lung segmentation is a predecessor technique to the development of automated Computer-Aided Diagnosis (CAD) system for Computed Tomography slices of the lungs. It is the  process of separating the lung region from other anatomical portion of the body in chest CT image.
1.2.1 Segmentation techniques
Medical image segmentation is one of the most challenging problems in health care industry and has been studied extensively in the last few decades. Research in segmentation methods has resulted in the development of new techniques for analysis of many different anatomical regions using image data acquired from a variety of modalities. Several image processing techniques such as thresholding, edge detection, region growing, and morphological operations have been proposed earlier for the effective segmentation of medical images. Each of the above approaches has their own advantages and limitations. However, manual segmentation  performed by domain experts is considered to be the most accurate but is time-consuming. Moreover, semiautomatic segmentation requires minimal user intervention since it is necessary to provide suitable inputs and parameters to facilitate the accuracy of segmentation. Since automatic segmentation does not require any user input, it is more complex in terms of computation and obtaining accurate results. In applications involving the processing of large number of images, automatic segmentation may be preferred for obtaining accurate results.
1.2.1.1 Fuzzy logic based segmentation
Fuzzy logic (FL) is a concept that assumes inexactness in practical situations. In reality, data obtained is not  perfect to the expectation during experimentation. The data can fit to its defined category or possibly to an adjacent category. This looseness of association of data to more than one category leads to the development of fuzzy logic.
International Journal of Applied Engineering Research, ISSN 0973-4562 Vol.9 No.27 (2014)© Research India Publications; httpwww.ripublication.comijaer.htm9612
 
This concept is very much used in control related applications in Engineering. FL can be combined with ANN/ Genetic Algorithm or other types of mathematical algorithms to achieve an input-output data mapping concept. Many researchers had used fuzzy related concepts in segmenting the CT lung image
1.2.1.2 Artificial Neural Network based segmentation
Artificial Neural Network (ANN) is the simulation of the functionality of human brain. The working of ANN is realized by implementing the mathematical representation of ANN through programming. The implementation of ANN for an application can be done by using one of the supervised / unsupervised / recurrent algorithms. ANN maps sets of input-output data. Two phases of implementation of ANN for an application is required. In phase-1, connection weights are updated by mapping input-output data. In phase-2, outputs are obtained when an input is given.
1.2.1.3 Pixel-Based Segmentation in CT
There are a number of semi-automated, pixel-based methods for segmenting CT images. In pixel–based methods, the initial process is to eliminate fat tissue and bones present in the CT lung images. The lung parenchyma has a very low density and low intensity of pixels in the CT scan. This property is utilized for the process of separating two lungs from the surrounding tissue. A technique commonly used in semiautomatic segmentation of CT images is gray-level thresholding. By knowing the X-ray attenuation of different tissues, it can be easily identified by differentiating the pixels within a specified range of gray-level. Histogram analysis is used to set a threshold values. The set threshold is combined with seeded region-growing so that the pixels are included in the segmentation region. Denison, 1986, used a two-dimensional (2-D) region growing in order to segment the lungs using a seed point which is provided by an operator for each image slices. A new methodology called edge-tracking has been applied by Hedlund, 1982, in order to segment the lung portion. In their method, the operator providing a seed point initiates the pixel to pixel search to the Region of Interest (ROI) of lung-pleural interface. The major drawback behind the pixel based segmentation is that the neighboring pixels are not considered while assigning segmentation to each and every pixel. Leader, 2003, proposed a fully-automated lung segmentation procedure for segmenting the lung region in computed tomography lung images. A pixel-value threshold along with two sets of classification rules was used for each and every slice. Features such as size, circularity, and location are extracted from the slices. In their work, first the background pixels were removed by preprocessing. The lung tissues are identified by pixel value thresholding. The initially segmented regions are refined in order to detect the airways and to separate the fused right and left lungs.
1.2.1.4 Thresholding based Segmentation in CT
 The thresholding based methods depends mostly on the brightness constant called threshold value of the  pixels in the original image. This thresholding based method works well in images which has a bimodal distribution. These threshold-based methods perform well in chest CT scans of patients with normal lungs. Ukil and Reinhardt (2005) proposed a 3D threshold-based based method for segmenting the lung images. They improved the work done by Hu et al (2001) by introducing a smoothing operator at the mediastinal area in order to guarantee the efficiency consistency among segmenting different slices. Their method increases the intra-subject over time taken for segmentation. The major drawback behind their method is the number of scans used in their method which was only 8. Sun et al (2006) presented a 3D method for the segmentation of lung region from thick slice of CT images. In their method, first a preprocessing was applied by using an anisotropic filter in order to improve the signal-to-noise ratio. A wavelet transform-based interpolation method was used to construct 3D volume of data. The lung regions were obtained by region growing using gray-value, homogeneity and gradient magnitude as input in these 3D volume data. The cavities present inside the resulting lung region were filled using morphological operation called closing. A noval method for segmenting the lung regions by combining threshold-based method with adaptive border marching was proposed by Pu et al (2008). They have analyzed their algorithm for 20 slices. The metrics used to evaluate their segmentation accuracy is volume-based over and under segmentation. Meng et al (2012) illustrated the performance of threshold-based lung segmentation algorithms in a large-scale robustness analysis. They have applied their proposed method on a set of 2768 chest CT scans which consists of 2292 slices. The slices are taken under various abnormalities and scanning positions. Automatic threshold-based lung segmentation was performed and visually checked in all the slices. Average error rate of 4.4% was obtained in 121 scans. The segmentation errors are due to disease such as interstitial lung disease, artifacts and from the corrupted Digital Imaging and
International Journal of Applied Engineering Research, ISSN 0973-4562 Vol.9 No.27 (2014)© Research India Publications; httpwww.ripublication.comijaer.htm9613

Reward Your Curiosity

Everything you want to read.
Anytime. Anywhere. Any device.
No Commitment. Cancel anytime.
576648e32a3d8b82ca71961b7a986505