Lung Cancer Detection Using Digital Image Processing Techniques: A Review

From last decade, lung cancer become sign of fear among the people all over the world. As a result, many countries generate funds and give invitation to many scholars to overcome on this disease. Many researchers proposed many solutions and challenges of different phases of computer aided system to detect the lung cancer in early stages and give the facts about the lung cancer. CV (Computer Vision) play vital role to prevent lung cancer. Since image processing is necessary for computer vision, further in medical image processing there are many technical steps which are necessary to improve the performance of medical diagnostic machines. Without such steps programmer is unable to achieve accuracy given by another author using specific algorithm or technique. In this paper we highlight such steps which are used by many author in pre-processing, segmentation and classification methods of lung cancer area detection. If pre-processing and segmentation process have some ambiguity than ultimately it effects on classification process. We discuss such factors briefly so that new researchers can easily understand the situation to work further in which direction.


INTRODUCTION
lung cancer namely SCLC (Small-Cells Lung Carcinoma) which is related to smoking of any types and the other one is NSCLC (Non-Small Cell Lung Carcinoma), 10-15% lung cancer diagnostics as SCLC while 85% cases of lung cancer are NSCLC [1]. Among multiple symptoms of lung cancer, some common symptoms are cough that get worse, chest pain, weight loss, shortness of breath, coughing of blood and weariness etc. In order to overcome the death rate due to lung cancer, one should adopt screening, chest radiograph (X-Ray), MRI (Magnetic Resonance Imaging)

Pre-Processing of Lung Cancer
Pre-processing is an improvement of the image data that suppresses unwanted distortions or enhances some image features for further processing.It is needed to minimizing the effects of distortion found in imaging device such as light fluctuation, to remove blueness and in the same time pre-processing is required to remove unwanted areas from the images and some time it is used for enhancing the image features like lines, boundaries and textures of image so that we can easily divide the contents of images in two parts, wanted and un-wanted contents of image.
For removing noise from the image, many researchers use different filtering techniques which depends on type of noise. In medical imaging all types of filtering techniques may be used depending on noise present in image [1]. Detail is given below: (a)

Image Enhancement
In the process of features extraction from the image it is necessary that all properties of an object present in the image should be clear. So, for growing the digital objects we need image enhancement [3]. Image enhancement is divided into two categories. First one is spatial domain and other is frequency domain.
In spatial domain, operations are performed on the pixel values directly so it is easy to understand and analysis.
While in frequency domain the methods is used to explain the analysis of signals and mathematical formulas with respect to frequency and function. Image enhancement is achieved when we are able to interpret and threshold the image into two parts, one is known as ROI (Region of Interest) and second is compliment of it.

Gabor Filer
Gabor filter is belonging to the linear filter mentioned in

Wiener Filter
Wiener filter is also linear filter belong to special domain.

Layer Separation
There are many imaging model like RGB (Red, Green, Blue)

HSV (Herpes Simplex Virus), HIS (Humane Society
International), YCbCr (Luminance; Chroma: Blue; Chroma: Red) and Lab model. In RGB model, RGB are also called color Channels. Sometime we take single channel which is consistent in different situation or distortion during the imaging process. The separation of a channel from other channels is called layer separation, which is key step to produce the satisfactory results [9].

Image Segmentation
The process of subdividing the image into two parts one is wanted and other is unwanted parts. Wanted part is also called ROI. In the case of lung cancer our target is to identify the tumor present in the Lung image ( Fig. 3(a-b)).
So tumor is our ROI and other part is unwanted area [11].
It is crucial task for machine to automatically detect the tumor because of variant texture properties of Lung tissue in abnormal region such as reûective of tumors or level of a cancer malignancy [12].

Region Growing
In region growing method we group the pixels or subparts of the image into the large region, means to say we analysis

Clustering Techniques
In

Edge Based Segmentation Methods
In edge based segmentation we try to find edges of digital objects found in an image on the basis of abruptly change found in intensity values of a pixel (Fig. 5). Many digital objects have cloudy and broken edges. To remove such effects from the image we usually performs morphological operations which are discussed in next section and linear filters to give more distance and remove distance which is closed to the digital object [16,19]. Because tumor has no regular shape and intensity range so it is challenging task for CAD (Computer Aided Design).

Morphological base Segmentation
Morphological base segmentation is a process that combine morphological operations such as extended

Template Matching
In template matching method we take a sample or template This initial vector is taken from shift vector obtained from local matching.

Marker-Controller Segmentation
To

Feature Extraction
Each well define object has some features or properties which helps us to identify that object easily. To automatically classify the objects, we need some features.
Collection of these features are called features vector.

Banalization
The banalization algorithm is also known as the threshold algorithm. The purpose of this algorithm is to find the suitable threshold [19]. By use of this threshold value the targeted image is divided into two parts foreground and background image. Foreground image is representing by white and background is representing by black color. This approach is based on the number of black pixels which are greater than white pixels if not then it means image is not normal.

Masking
Masking is also called filtering the word "filtering" has been borrowed from the frequency domain. Filters are classified as: Low-Pass (i.e. preserve low frequencies) High-Pass (i.e. preserve high frequencies) Band-Pass (i.e. preserve frequencies within a band) Band-Reject (i.e. reject frequencies within a band) [20].
In masking approach, we take usually neighborhood pixels of CT scan images of lung cancer with effected areas and then calculate the statistical and structural features which are used to classify Lung cancer images in early stage and normal images as well.

Classification
Human brain is a great example of ability to learn and classify the different objects and things without any error.

Multi-Label Classification
Multi-label classification in concerned with the classiûcation about information for different class labels. MLC (Multi-Label Classification) is a major research area in the machine learning community and ûnds application in several domains. MLC problem by adopting a groupsparsity regularize to select common subsets of relevant features from different cellular compartments. In addition, we also add a cell structural correlation regularized Laplacian term, which utilizes the prior biological structural information to capture the intrinsic dependency among different cellular compartments [26].

CONCLUSION
An effort to diagnose the cancer in the lung using digital image processing techniques. Initially the CT scan/MRI captured the image and processed then cancer or tumor region is detected and diagnosis from the original image.
In pre-processing phase, number of techniques are being used for filter the distortion of the images, for example Weiner Filter, Fast Fourier Transformation etc. After preprocessing, image segmentation phase should be done through Marker-Controlled Watershed and Watershed Segmentation. The next phase is features extraction is done by Area, Perimeter, and Eccentricity etc. These features help to detect the lung cancer. In classification, SVM (Support Vector Machine) and CNN (Convolutional Neural Network) are used for classification of normal and abnormal area in early diagnoses.

FUTURE WORK
In future researchers will try to build Image Processing Techniques on X-Ray, PET images for more accuracy. For the classification KNN or fuzzy, logic k/c-mean and clustering may be used. Furthermore, contrast between X-Ray and CT scan images for better result for early lung cancer diagnose.