Short on time?

Get essay writing help

Lung Cancer Detection and Classification Using SVM

Words: 1996
Pages: 4
This essay sample was donated by a student to help the academic community. Papers provided by EduBirdie writers usually outdo students' samples.


Image processing techniques are widely used in several medical problems for image enhancement in the detection phase to support early medical treatment. In this research, we aim to improve quality and accuracy of early detection of lung cancer through a combination of image processing techniques and machine learning. The Cancer Imaging Archive (TCIA) dataset has been used for training and testing purpose where DICOM is the primary format used for image storage. Classification is done using SVM (Support Vector Machine) classifier which identifies whether the CT image is cancerous and non-cancerous. Before training the classifier, we are performing image processing techniques on CT images such as converting the image into HU (Hounsfield Unit) scale to get the binary image of lungs followed by nodule segmentation where nodules are detected within the lungs. Further, features are extracted from CT images using GLCM (Gray Level Co-Occurrence Matrix) .The enhanced image is then given to the classifier to provide accurate results. As an enhancement, we have also performed Early Stage Detection for reducing the growing cancer burden. MATLAB image processing toolbox based has been used for the implementation on the CT scan images.

Keywords—Nodules Detection, Image Processing, Classification, Machine Learning


Cancer is one of the most dangerous diseases that causes death. Lung cancer has become one of the most common causes in both men and women. A large number of people die every year due to lung cancer. The disease has different stages whereby it starts from the small tissue tissues and spreads throughout the different parts of the body. According to the American Cancer Society [9] the cases of lung cancer increases very rapidly and almost 14% newly diagnosed cancers are lung cancer and also the main cause of cancer death worldwide. The previous study of diagnosis showed that most of the lung cancer patients belong to the age of 60 years. The process of early detection of cancer plays an important role to prevent cancer cells from multiplying and spreading. Although CT scan imaging is best imaging technique that are reliable for lung cancer diagnosis because it can disclose every suspected and unsuspected lung cancer nodule in the medical field. However, the variance of intensity in CT images and anatomical structure misjudgments by doctors and radiologists might cause difficulty to interpret and identify cancer from CT scan images. Therefore computer-aided diagnosis can be helpful to assist radiologists and doctors to identify cancerous cells accurately.

Previous research have been conducted for analyzing lung cancer, where the detection of lung cancer with image processing techniques provides good results and accuracy. In this study, we proposed different methods in analyzing lung cancer using image processing techniques along with machine learning algorithm for the classification of the CT scan images whether it is cancerous or non-cancerous. The main aim of this research is to provide accurate results by the means of the machine learning algorithm.

Literature Review

Several researchers have proposed and implemented the detection of lung cancer using different approaches of image processing and machine learning. According to our survey, Aggarwal, Furquan and Kalra [2] proposed a model that provides classification between nodules and normal lung anatomy structure. The method extracts geometrical, statistical and gray level characteristics. LDA is used as a classifier and optimal thresholding for segmentation. The system has 84% accuracy, 97.14% sensitivity and 53.33%specificty. Although the system detects the cancer nodule, its accuracy is still unacceptable. No any machine learning techniques has been used to classify and simple segmentation techniques is used. Ignatius and joseph [6] developed a system using watershed segmentation. In preprocessing, it uses a Gabor filter to enhance the image quality. The accuracy of the proposed is 90% which is comparatively higher than the model with segmentation using neural fuzzy model and region growing method. As a limitation, it does not classify cancer as benign and malignant. Gonzalez and Ponomaryvo [7] proposed a system that classifies lung cancer as benign or malignant. The system uses the priori information and HU to calculate the Region of Interest. Shape features like area, eccentricity, fractal dimension and textual features are extracted to train and classify the support vector machine to identify whether the nodule is cancerous our non-cancerous. Gonzalez and Ponomaryvo is the current best model. However, it does not make any effort for early detection of lung cancer which will improve the overall chances of survival rate for the patient and thus leads to one of the limitations of this study. Also no such filtering techniques are used which would remove the noise and improve the accuracy of the system.

Proposed System

Figure 1 shows a general description of lung cancer detection system that contains basic stages:

0. Collecting Dataset

Image Processing(Median Filter)

Converting Image to HU scale

Segmentation(Lungs and Nodules)

Feature Extraction


Stage Detection

CT scan lung images

Image Denoising using Median Filter

Converting Image to HU Scale

Segmentation (Lungs and Nodules)

SVM Classification

Benign or Malignant

Fig-1:Flowchart of Proposed System

A. Collecting Dataset

The Lung Cancer Image Dataset [8] has been chosen from TCIA(The Cancer Imaging Archive) which is an online CT scan image dataset publicly available for researchers in the field of digital image processing. Images in the dataset are in the DICOM format with a size 512*512 pixel. In TCIA dataset, we have used 20% data for testing and the remaining 80% of the dataset is used for training the SVM classifier.

B. Image Processing

Smoothing is an image processing technique used in order to reduce noise in an image to produce clearer image. Most of the techniques are based on low pass linear filters. It is mostly based on the averaging technique of the input image. To perform a smoothing operation we will apply a filter to our image. The most common type of filter is the median filter which is used in our proposed system.

C. Converting image to HU Scale

Hounsfield Unit is the unit of measurement used in CT scan images which is a measure of radiodensity. CT scanners are carefully calibrated to accurately measure this.HU scale is a linear transformation of the original linear attenuation coefficient measurement into one in which the radiodensity of distilled water at standard pressure and temperature (STP) is defined as zero HU, while the radiodensity of air at STP is defined as -1000 HU. The HU value is therefore given by

Fig 2. Hounsfield Formula


Hounsfield Unit








-100 to -50







+30 to +45


+10 to +40

Grey matter

+37 to +45

White matter

+20 to +30


+40 to +60

Soft Tissue, Contrast

+100 to +200


+700(cancerous bone) to +3000(cortical bone)

Save your time!
We can take care of your essay
  • Proper editing and formatting
  • Free revision, title page, and bibliography
  • Flexible prices and money-back guarantee
Place Order

By default however, the returned values are not in this unit. But this can be fixed. Some scanners have cylindrical scanning bounds, but the output image is square. The pixels that fall outside of these bounds get the fixed value -2000. The first step is setting these values to 0, which currently corresponds to air. Next we convert our image back to HU units, by multiplying with the rescale slope and adding the intercept.


D. Segmentation

Segmentation is a process that splits the image into its constituent regions or objects. Segmentation is usually used to trace objects and borders such as lines, curves, etc. in images. The main objective of segmentation is to simplify and change the representation of the image into something that is more significant and easier to examine.

In our proposed system, Segmentation involves two parts which are Lung Segmentation and Nodules Segmentation.

Lung Segmentation is implemented through HU Scale where once the image is converted into HU Scale, we get segmented binary image of lungs. As a precautionary measure, we have also used active contour method to trace the lungs part properly so as to improve the accuracy. Further in Nodules Segmentation, nodules are detected within the binary image by using Morphological Operations such as dilation and erosion. Dilation and Erosion are often used in combination for specific image preprocessing applications such as filing holes or removing small objects. As a result of which, we get a segmented nodule image given to feature extraction to perform textual analysis on the image.

Fig 3. Original Image

Fig 4. Lungs Segmented Image

[image: D:FLung Cancer Detection with stagesModule 2DICOMSTraining SetMask2.bmp]

Fig 5. Nodule Segmented Image

E. Feature Extraction

Feature Extraction stage is a crucial stage for the Computer-Aided Diagnosis (CAD) system. It uses different methods and algorithms for feature extraction from the segmented image. The extracted image can be classified as either cancerous or non-cancerous using texture properties. We have used HARALICK GLCM (Gray Level Co-Occurrence Matrix) for texture feature extraction from CT scan images. The GCLM is a tabulation of how often different combinations of pixel brightness values (gray levels) occur in an image. Firstly we create a gray-level co-occurrence matrix from images in MATLAB.

Some of the features are extracted using this method are:

  • Contrast
  • Entropy
  • Energy
  • Homogeneity
  • Correlation

Then these features are used for tumor classification.

F. Classification

This stage classifies the detected nodule as malignant of benign. Support Vector Machine (SVM) is used as a classifier. SVM is supervised machine learning algorithm which defines the function that classifies data into two classes. In our proposed system, we have defined two classes as cancerous or non-cancerous. SVM is a binary classification method that takes as input labeled data from two classes and outputs a model file for classifying unknown or known data into one of two classes.

Fig 5. SVM Classifier

Training an SVM involves feeding known data to the SVM along with previously known decision values, thus forming a training set. It is from the training set that an SVM gets its intelligence to classify unknown data.





Output Value





























G. Stage Detection

Early Detection represents one of the most promising approaches for reducing the growing cancer burden. The purpose of early detection is that it will identify cancer while still localized and curable, preventing not only mortality but also reducing costs.

Future work

In the future scope, the accuracy of the system can be improved if training is performed by using a very large image database. Similarly, we have planned to use MRI scan images to detect different types of lung cancer or other cancers using the same approach.

V. Conclusion

The current best system does not classify into different stages of cancer such as stage I, II, III or IV which results into non-satisfactory results. Therefore new system is proposed. The proposed system is successfully able to classify both benign and malignant tumors more accurately. It is also possible for early-stage detection of lung cancer by using this system, thereby overcoming the limitations of the existing system. SVM can handle better complex classifications as compared to KNN classifiers.

1. References

  1. Xiuhua,G., Tao, S., & Zhigang, L.(2011) “Prediction Models for Malignant Pulmonary Nodules Based-on Texture Features of CT Image.” In Theory and Applications of CT Imaging and Analysis. DOI: 10.5772/14766.
  2. Aggarwal, T., Furqan, A., & Kalra, K. (2015) “Feature extraction and LDA based classification of lung nodules in chest CT scan images.” 2015 International Conference On Advances In Computing, Communications And Informatics (ICACCI), DOI: 10.1109/ICACCI.2015.7275773.
  3. Jin, X., Zhang, Y., & Jin, Q. (2016) “Pulmonary Nodule Detection Based on CT Images Using Convolution Neural Network.” 2016 9Th International Symposium On Computational Intelligence And Design (ISCID). DOI:10.1109/ISCID.2016.1053.
  4. Sangamithraa, P., & Govindaraju, S. (2016) “Lung tumour detection and classification using EK-Mean clustering.” 2016 International Conference On Wireless Communications, Signal Processing And Networking (Wispnet). DOI:10.1109/WiSPNET.2016.7566533.
  5. Roy, T., Sirohi, N., & Patle, A. (2015) “Classification of lung image and nodule detection using fuzzy inference system.” International Conference On Computing, Communication & Automation. DOI: 10.1109/CCAA.2015.7148560.
  6. Ignatious, S., & Joseph, R. (2015) “Computer-aided lung cancer detection system.” 2015 Global Conference On Communication Technologies (GCCT), DOI: 10.1109/GCCT.2015.7342723.
  7. Rendon-Gonzalez, E., & Ponomaryov, V. (2016) “Automatic Lung nodule segmentation and classification in CT images based on SVM.” 2016 9Th International Kharkiv Symposium On Physics And Engineering Of Microwaves, Millimeter And Submillimeter Waves (MSMW). DOI: 10.1109/MSMW.2016.7537995.
  8. The Cancer Imaging Archive(TCIA) Dataset
  9. American Cancer Society. Costs of Cancer, 2002 [online], (cited 25 Jan 2003),
  10. Sruthi Ignatious, Robin Joseph, Jisha John Dept. of Computer Science and Engineering, Mar Baselios college of Engineering and Technology Trivandrum, Dr. Anil Prahladan Dept. of Imageology, Regional Cancer Centre Trivandrum, ” Computer-Aided Lung Cancer Detection and Tumor Staging in CT image using Image Processing” India International Journal of Computer Applications (0975 – 8887) Volume 128 – No.7, October 2015

Make sure you submit a unique essay

Our writers will provide you with an essay sample written from scratch: any topic, any deadline, any instructions.

Cite this Page

Lung Cancer Detection and Classification Using SVM. (2022, March 18). Edubirdie. Retrieved September 28, 2023, from
“Lung Cancer Detection and Classification Using SVM.” Edubirdie, 18 Mar. 2022,
Lung Cancer Detection and Classification Using SVM. [online]. Available at: <> [Accessed 28 Sept. 2023].
Lung Cancer Detection and Classification Using SVM [Internet]. Edubirdie. 2022 Mar 18 [cited 2023 Sept 28]. Available from:
Join 100k satisfied students
  • Get original paper written according to your instructions
  • Save time for what matters most
hire writer

Fair Use Policy

EduBirdie considers academic integrity to be the essential part of the learning process and does not support any violation of the academic standards. Should you have any questions regarding our Fair Use Policy or become aware of any violations, please do not hesitate to contact us via

Check it out!
search Stuck on your essay?

We are here 24/7 to write your paper in as fast as 3 hours.