SlideShare a Scribd company logo
Lung Nodule Diagnosis from CT
Images Based on Ensemble Learning
Mohammad Hossein Fazel Zarandi
Farzad Vasheghani Farahani
Abbas Ahmadi
2015 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology
Niagara Falls, Canada | August 12 - 15
Topics
• Objective
• Procedure
• Background
• Proposed system
• Contribution
• Conclusion
• References
Objective
The purpose of this study is to “design a CAD system for diagnosis
of lung cancer” which is:
• Helping physicians as a second opinion
• Fast and accurate for early diagnosis of this disease
• Based on processing CT images and image analysis techniques
• Using ensemble of classifiers
Procedure
How to achieve the desired objectives?
Background
About lung cancer
• Cancer is one of top mortal diseases in the world [1] like
heart diseases and diabetes.
• Lung cancer is the most leading cause of deaths in both
women and men (27.2% of all cancer deaths) [2].
• Lung cancer is caused by an uncontrollable irregular
growth of cells in lung tissue.
• Lung tissue abnormalities that are roughly spherical with
round opacity and a diameter of up to 30mm [3] are
known as lung nodules.
Computer-aided diagnosis (CAD) systems
• CAD systems Are very helpful for radiologists in the detection and
diagnosing abnormalities earlier and faster [4].
• Indeed, computer-aided
diagnosis systems act as
a second opinion for
radiologists before
suggesting a biopsy test
[5].
Machine learning methods
• Ensemble learners combine decisions
of base classifiers to form an
integrated output [8].
• Machine learning methods have become one of the dominant
approaches in CAD for medical imaging [6].
• One of the four pioneer branches in the field of machine learning is
ensemble based systems [7].
Proposed System
Systematic overview
Image
Processing
Classification
Lung CT dataset
• 60 CT images including both men and women collected by Lung
Image Database Consortium (LIDC).
• Lung lesions in these images are marked by up to four
radiologists in VIA group.
• All the lung CT images are in the
format of DICOM with the size of
512*512.
Image processing phase
Image processing module in this model consists of 4 main steps:
Preprocessing
Segmentation
Post
processing
Feature
extraction
Raw images
Extracted
features
1) Preprocessing
• Image preprocessing is a process
which eliminates primary noise and
image distortion, and also enhances
important features exists in CT
images.
• By using some image enhancement
methods including median filter
and histogram equalization.
2) Segmentation
a) Background elimination and other additional parts such as bones to
extract lung tissue and region of interest (ROI) in the lung image.
b) Convert the image to black and white in order to prepare it for
boundary tracing.
Method: Region growing
Method: Thresholding
2) Segmentation (Cont.)
Background elimination
using region growing
Convert to black and
white using thresholding
3) Post processing
• Eliminate unwanted objects (small or
large) inside and outside the binary
image of lungs.
• Join the detected object in the image
together by filling in the gaps between
them and by smoothing their outer
edges.
• Fill image regions and holes of the
detected objects.
• Find boundaries of the objects by
concentrating only on exterior
boundaries.
4) Feature extraction
In order to classify suspected objects as nodule or non-nodule,
further working is needed to extract specific features from raw
images.
Feature Definition
Roundness
4𝐴
𝜋𝐿2
Circularity
4𝜋𝐴
𝑃2
Compactness 𝑃2
4𝜋𝐴
Ellipticity 𝜋𝐿2
2𝐴
Eccentricity −
• In this study, five morphological
features are extracted for each
suspected objects from CT images.
• P, A and L denote the perimeter, area
and maximum diameter of the
suspected objects, respectively.
Classification phase
One of the suitable methods to improve the accuracy of
classification is the use of multiple classifier systems.
Multiple classifier systems generally
consist of two main parts [9]:
• Creation an ensemble
• Combination of class labels
1) Creating an ensemble
Three different classifiers including Multilayer perceptron (MLP),
K-Nearest neighbor (KNN), and Support vector machines (SVM)
work on numerical data obtained from feature vectors.
Note: The extracted features from suspected
objects are considered as inputs to the ensemble
system and they are normalized between zero and
one. Hence, the feature set consists of five
extracted features from the previous section and a
binary target feature that defines suspected
objects are nodule or not (0 or 1).
2) Combining outputs
Each base classifier has its own opinion and identifies suspected
objects as nodule or non-nodule.
Majority voting techniques is used to combine the results of base
classifiers.
Final output of
ensemble system
System Performance Measurements
Accuracy (Acc), Sensitivity (Sn) , Specificity (Sp), G-mean , Precision
(Prc) and F-measure of the proposed system:
Proposed
method
System Performance Measurements
Acc Sn Sp G-mean Prc F-measure
MLP 90.41 73.55 94.68 83.45 77.76 75.60
KNN 91.20 81.76 93.59 87.48 76.33 78.96
SVM 90.60 73.44 94.94 83.50 78.59 75.93
Majority
Voting
94.65 84.16 97.30 90.49 88.76 86.40
Confusion matrix
MLP
Actual Predicted
Negative Positive
Negative 75.56 % 4.25 %
Positive 5.34 % 14.85 %
KNN
Actual Predicted
Negative Positive
Negative 74.69 % 5.12 %
Positive 3.68 % 16.51 %
SVM
Actual Predicted
Negative Positive
Negative 75.77 % 4.04 %
Positive 5.36 % 14.83 %
Proposed method
Actual Predicted
Negative Positive
Negative 77.66 % 2.15 %
Positive 3.20 % 16.99 %
Contribution
The main contribution of this paper is to:
present a classifier ensemble system that uses image processing
and data mining techniques for lung nodule identification.
• Feature extraction of suspected objects in CT images is the main
goal of image processing phase.
• In data mining phase, there are three different classifiers including
MLP, KNN and SVM that work on extracted features to identify lung
objects as nodule or non-nodule in the form of group decision-
making.
Conclusion
• In present work, the manner of a CAD system in diagnosis process is
simulated.
• Our proposed CAD system is a highly promising method, providing high
performance in the identification of pulmonary nodules from CT images.
• For the sake of checking the effectiveness of the proposed hybrid
intelligent method, 5-fold cross-validation approach was applied and the
program ran for 10 times.
• Performance measurements show that the system could be used as a
second opinion for physicians in the diagnosis of lung cancer from CT
images.
References
1. “WHO | The top 10 causes of death.”
2. “Cancer of the Lung and Bronchus - SEER Stat Fact Sheets.” [Online]. Available:
http://seer.cancer.gov/statfacts/html/lungb.html. [Accessed: 13-Feb-2015].
3. J. H. M. Austin, P. Naidich, L. Muller, M. Hansell, and A. Zerhouni, “Glossary of terms for CT of the lungs:
recommendations of the Nomenclature Committee of the Fleischner Society,” Radiology, vol. 200, no. 2, pp.
327–331, 1996.
4. R. N. Strickland, “Tumor detection in non stationary backgrounds,” IEEE Trans. Med. Imaging, vol. 13, no. 3, pp.
491–9, Jan. 1994.
5. S. B. Lo, S. A. Lou, J. Lin, M. T. Freedman, M. V Chien, and S. K. Mun, “Artificial Convolution Neural Network
Techniques and Applications for Lung Nodule Detection,” Med. Imaging, IEEE Trans., vol. 14, no. 4, pp. 711–718,
1995.
6. S. L. a. Lee, a. Z. Kouzani, and E. J. Hu, “Automated identification of lung nodules,” 2008 IEEE 10th Work.
Multimed. Signal Process., pp. 497–502, Oct. 2008.
7. T. G. Dietterich, “Machine-Learning Research,” AI Mag., vol. 18, no. 4, pp. 97–136, 1997.
8. S. L. a Lee, a Z. Kouzani, and E. J. Hu, “Random forest based lung nodule classification aided by clustering.,”
Comput. Med. Imaging Graph., vol. 34, no. 7, pp. 535–42, Oct. 2010.
9. R. Polikar, “Ensemble Based Systems in Decision Making,” Circuits Syst. Mag. IEEE, vol. 6, no. 3, pp. 21–45,
2006.
Thanks for listening …

More Related Content

What's hot

Head and neck; brachytherapy.pptx final
Head and neck;  brachytherapy.pptx finalHead and neck;  brachytherapy.pptx final
Head and neck; brachytherapy.pptx final
pgclubrcc
 
Stereotactic Radio-Surgery/Therapy (SRS/SRT)
 Stereotactic Radio-Surgery/Therapy (SRS/SRT) Stereotactic Radio-Surgery/Therapy (SRS/SRT)
Stereotactic Radio-Surgery/Therapy (SRS/SRT)
Aaditya Sinha
 
Medical image analysis
Medical image analysisMedical image analysis
Medical image analysis
Aboul Ella Hassanien
 
Artificial intelligence in radiology
Artificial intelligence in radiologyArtificial intelligence in radiology
Artificial intelligence in radiology
Dev Lakhera
 
Dose reduction in MDCT . Daniel J.P , Khorfakhan hospital . UAE
Dose reduction in MDCT . Daniel J.P , Khorfakhan hospital . UAEDose reduction in MDCT . Daniel J.P , Khorfakhan hospital . UAE
Dose reduction in MDCT . Daniel J.P , Khorfakhan hospital . UAE
haijaypee_dan
 
HRCT CHEST/TEMPORAL BONE PROTOCOL.pptx
HRCT CHEST/TEMPORAL BONE PROTOCOL.pptxHRCT CHEST/TEMPORAL BONE PROTOCOL.pptx
HRCT CHEST/TEMPORAL BONE PROTOCOL.pptx
SAMEER AHMAD GANAIE
 
Ct virtual endoscopy 1
Ct virtual endoscopy 1Ct virtual endoscopy 1
Ct virtual endoscopy 1
DeepikaHamav
 
The principles of head and neck PET/CT
The principles of head and neck PET/CTThe principles of head and neck PET/CT
The principles of head and neck PET/CT
Dr- Mustafa Ahmed Alazam
 
Radiomics and Deep Learning for Lung Cancer Screening
Radiomics and Deep Learning for Lung Cancer ScreeningRadiomics and Deep Learning for Lung Cancer Screening
Radiomics and Deep Learning for Lung Cancer Screening
Wookjin Choi
 
THE RATIONALE AND BENEFITS OF IGRT
THE RATIONALE AND BENEFITS OF IGRTTHE RATIONALE AND BENEFITS OF IGRT
THE RATIONALE AND BENEFITS OF IGRTMelissa McClement
 
Imrt delivery
Imrt deliveryImrt delivery
Imrt delivery
makhhi
 
CT SCAN - 3 Marks - QUESTION AND ANSWERS
CT SCAN - 3 Marks - QUESTION AND ANSWERSCT SCAN - 3 Marks - QUESTION AND ANSWERS
CT SCAN - 3 Marks - QUESTION AND ANSWERS
Ganesan Yogananthem
 
parsport trial ppt
parsport trial pptparsport trial ppt
parsport trial ppt
Gaurav Kumar
 
Patient positional correction stategies in radiotherapy
Patient positional correction stategies   in radiotherapyPatient positional correction stategies   in radiotherapy
Patient positional correction stategies in radiotherapy
Biplab Sarkar
 
Basic principle of ct and ct generations
Basic principle of ct and ct generationsBasic principle of ct and ct generations
Basic principle of ct and ct generations
TarunGoyal66
 
Lec16: Medical Image Registration (Advanced): Deformable Registration
Lec16: Medical Image Registration (Advanced): Deformable RegistrationLec16: Medical Image Registration (Advanced): Deformable Registration
Lec16: Medical Image Registration (Advanced): Deformable Registration
Ulaş Bağcı
 
Artificial intelligence-in-radiology
Artificial intelligence-in-radiologyArtificial intelligence-in-radiology
Artificial intelligence-in-radiology
Vrishit Saraswat
 
Dose reduction technique in ct scan
Dose reduction technique in ct scanDose reduction technique in ct scan
Dose reduction technique in ct scan
Mohd Aiman Azmardi
 
CNN Machine learning DeepLearning
CNN Machine learning DeepLearningCNN Machine learning DeepLearning
CNN Machine learning DeepLearning
Abhishek Sharma
 

What's hot (20)

Head and neck; brachytherapy.pptx final
Head and neck;  brachytherapy.pptx finalHead and neck;  brachytherapy.pptx final
Head and neck; brachytherapy.pptx final
 
Stereotactic Radio-Surgery/Therapy (SRS/SRT)
 Stereotactic Radio-Surgery/Therapy (SRS/SRT) Stereotactic Radio-Surgery/Therapy (SRS/SRT)
Stereotactic Radio-Surgery/Therapy (SRS/SRT)
 
Radiation Protection
Radiation ProtectionRadiation Protection
Radiation Protection
 
Medical image analysis
Medical image analysisMedical image analysis
Medical image analysis
 
Artificial intelligence in radiology
Artificial intelligence in radiologyArtificial intelligence in radiology
Artificial intelligence in radiology
 
Dose reduction in MDCT . Daniel J.P , Khorfakhan hospital . UAE
Dose reduction in MDCT . Daniel J.P , Khorfakhan hospital . UAEDose reduction in MDCT . Daniel J.P , Khorfakhan hospital . UAE
Dose reduction in MDCT . Daniel J.P , Khorfakhan hospital . UAE
 
HRCT CHEST/TEMPORAL BONE PROTOCOL.pptx
HRCT CHEST/TEMPORAL BONE PROTOCOL.pptxHRCT CHEST/TEMPORAL BONE PROTOCOL.pptx
HRCT CHEST/TEMPORAL BONE PROTOCOL.pptx
 
Ct virtual endoscopy 1
Ct virtual endoscopy 1Ct virtual endoscopy 1
Ct virtual endoscopy 1
 
The principles of head and neck PET/CT
The principles of head and neck PET/CTThe principles of head and neck PET/CT
The principles of head and neck PET/CT
 
Radiomics and Deep Learning for Lung Cancer Screening
Radiomics and Deep Learning for Lung Cancer ScreeningRadiomics and Deep Learning for Lung Cancer Screening
Radiomics and Deep Learning for Lung Cancer Screening
 
THE RATIONALE AND BENEFITS OF IGRT
THE RATIONALE AND BENEFITS OF IGRTTHE RATIONALE AND BENEFITS OF IGRT
THE RATIONALE AND BENEFITS OF IGRT
 
Imrt delivery
Imrt deliveryImrt delivery
Imrt delivery
 
CT SCAN - 3 Marks - QUESTION AND ANSWERS
CT SCAN - 3 Marks - QUESTION AND ANSWERSCT SCAN - 3 Marks - QUESTION AND ANSWERS
CT SCAN - 3 Marks - QUESTION AND ANSWERS
 
parsport trial ppt
parsport trial pptparsport trial ppt
parsport trial ppt
 
Patient positional correction stategies in radiotherapy
Patient positional correction stategies   in radiotherapyPatient positional correction stategies   in radiotherapy
Patient positional correction stategies in radiotherapy
 
Basic principle of ct and ct generations
Basic principle of ct and ct generationsBasic principle of ct and ct generations
Basic principle of ct and ct generations
 
Lec16: Medical Image Registration (Advanced): Deformable Registration
Lec16: Medical Image Registration (Advanced): Deformable RegistrationLec16: Medical Image Registration (Advanced): Deformable Registration
Lec16: Medical Image Registration (Advanced): Deformable Registration
 
Artificial intelligence-in-radiology
Artificial intelligence-in-radiologyArtificial intelligence-in-radiology
Artificial intelligence-in-radiology
 
Dose reduction technique in ct scan
Dose reduction technique in ct scanDose reduction technique in ct scan
Dose reduction technique in ct scan
 
CNN Machine learning DeepLearning
CNN Machine learning DeepLearningCNN Machine learning DeepLearning
CNN Machine learning DeepLearning
 

Similar to Lung nodule diagnosis from CT images based on ensemble learning

Iaetsd classification of lung tumour using
Iaetsd classification of lung tumour usingIaetsd classification of lung tumour using
Iaetsd classification of lung tumour using
Iaetsd Iaetsd
 
Early Detection of Lung Cancer Using Neural Network Techniques
Early Detection of Lung Cancer Using Neural Network TechniquesEarly Detection of Lung Cancer Using Neural Network Techniques
Early Detection of Lung Cancer Using Neural Network Techniques
IJERA Editor
 
Prediction of Lung Cancer Using Image Processing Techniques: A Review
Prediction of Lung Cancer Using Image Processing Techniques: A ReviewPrediction of Lung Cancer Using Image Processing Techniques: A Review
Prediction of Lung Cancer Using Image Processing Techniques: A Review
aciijournal
 
Prediction of Lung Cancer Using Image Processing Techniques: A Review
Prediction of Lung Cancer Using Image Processing Techniques: A ReviewPrediction of Lung Cancer Using Image Processing Techniques: A Review
Prediction of Lung Cancer Using Image Processing Techniques: A Review
aciijournal
 
Prediction of lung cancer using image
Prediction of lung cancer using imagePrediction of lung cancer using image
Prediction of lung cancer using image
aciijournal
 
Development of Computational Tool for Lung Cancer Prediction Using Data Mining
Development of Computational Tool for Lung Cancer Prediction Using Data MiningDevelopment of Computational Tool for Lung Cancer Prediction Using Data Mining
Development of Computational Tool for Lung Cancer Prediction Using Data Mining
Editor IJCATR
 
NCERTE-2016_ResearchPaper
NCERTE-2016_ResearchPaperNCERTE-2016_ResearchPaper
NCERTE-2016_ResearchPaperPanth Shah
 
IRJET- Lung Segmentation and Nodule Detection based on CT Images using Image ...
IRJET- Lung Segmentation and Nodule Detection based on CT Images using Image ...IRJET- Lung Segmentation and Nodule Detection based on CT Images using Image ...
IRJET- Lung Segmentation and Nodule Detection based on CT Images using Image ...
IRJET Journal
 
P.Surendar - VIVA PPT.pptx
P.Surendar - VIVA PPT.pptxP.Surendar - VIVA PPT.pptx
P.Surendar - VIVA PPT.pptx
surendar1989
 
Lung Conditions Prognosis Using CNN Model.pptx
Lung Conditions Prognosis Using CNN Model.pptxLung Conditions Prognosis Using CNN Model.pptx
Lung Conditions Prognosis Using CNN Model.pptx
DrIndrajeetKumar
 
BRAIN TUMOR DETECTION
BRAIN TUMOR DETECTIONBRAIN TUMOR DETECTION
BRAIN TUMOR DETECTION
IRJET Journal
 
computer aided detection of pulmonary nodules in ct scans
computer aided detection of pulmonary nodules in ct scanscomputer aided detection of pulmonary nodules in ct scans
computer aided detection of pulmonary nodules in ct scans
Wookjin Choi
 
Journal article
Journal articleJournal article
Journal article
graphicdesigner79
 
research on journaling
research on journalingresearch on journaling
research on journaling
graphicdesigner79
 
CLASSIFICATION OF LUNGS IMAGES FOR DETECTING NODULES USING MACHINE LEARNING
CLASSIFICATION OF LUNGS IMAGES FOR DETECTING NODULES USING MACHINE LEARNING CLASSIFICATION OF LUNGS IMAGES FOR DETECTING NODULES USING MACHINE LEARNING
CLASSIFICATION OF LUNGS IMAGES FOR DETECTING NODULES USING MACHINE LEARNING
sipij
 
Classification of Lungs Images for Detecting Nodules using Machine Learning
Classification of Lungs Images for Detecting Nodules using Machine LearningClassification of Lungs Images for Detecting Nodules using Machine Learning
Classification of Lungs Images for Detecting Nodules using Machine Learning
sipij
 
Statistical Feature-based Neural Network Approach for the Detection of Lung C...
Statistical Feature-based Neural Network Approach for the Detection of Lung C...Statistical Feature-based Neural Network Approach for the Detection of Lung C...
Statistical Feature-based Neural Network Approach for the Detection of Lung C...
CSCJournals
 
A REVIEW PAPER ON PULMONARY NODULE DETECTION
A REVIEW PAPER ON PULMONARY NODULE DETECTIONA REVIEW PAPER ON PULMONARY NODULE DETECTION
A REVIEW PAPER ON PULMONARY NODULE DETECTION
IRJET Journal
 
Automatic detection of tb
Automatic detection of tbAutomatic detection of tb
Automatic detection of tb
Shyama Menon
 
Multiple Analysis of Brain Tumor Detection Based on FCM
Multiple Analysis of Brain Tumor Detection Based on FCMMultiple Analysis of Brain Tumor Detection Based on FCM
Multiple Analysis of Brain Tumor Detection Based on FCM
IRJET Journal
 

Similar to Lung nodule diagnosis from CT images based on ensemble learning (20)

Iaetsd classification of lung tumour using
Iaetsd classification of lung tumour usingIaetsd classification of lung tumour using
Iaetsd classification of lung tumour using
 
Early Detection of Lung Cancer Using Neural Network Techniques
Early Detection of Lung Cancer Using Neural Network TechniquesEarly Detection of Lung Cancer Using Neural Network Techniques
Early Detection of Lung Cancer Using Neural Network Techniques
 
Prediction of Lung Cancer Using Image Processing Techniques: A Review
Prediction of Lung Cancer Using Image Processing Techniques: A ReviewPrediction of Lung Cancer Using Image Processing Techniques: A Review
Prediction of Lung Cancer Using Image Processing Techniques: A Review
 
Prediction of Lung Cancer Using Image Processing Techniques: A Review
Prediction of Lung Cancer Using Image Processing Techniques: A ReviewPrediction of Lung Cancer Using Image Processing Techniques: A Review
Prediction of Lung Cancer Using Image Processing Techniques: A Review
 
Prediction of lung cancer using image
Prediction of lung cancer using imagePrediction of lung cancer using image
Prediction of lung cancer using image
 
Development of Computational Tool for Lung Cancer Prediction Using Data Mining
Development of Computational Tool for Lung Cancer Prediction Using Data MiningDevelopment of Computational Tool for Lung Cancer Prediction Using Data Mining
Development of Computational Tool for Lung Cancer Prediction Using Data Mining
 
NCERTE-2016_ResearchPaper
NCERTE-2016_ResearchPaperNCERTE-2016_ResearchPaper
NCERTE-2016_ResearchPaper
 
IRJET- Lung Segmentation and Nodule Detection based on CT Images using Image ...
IRJET- Lung Segmentation and Nodule Detection based on CT Images using Image ...IRJET- Lung Segmentation and Nodule Detection based on CT Images using Image ...
IRJET- Lung Segmentation and Nodule Detection based on CT Images using Image ...
 
P.Surendar - VIVA PPT.pptx
P.Surendar - VIVA PPT.pptxP.Surendar - VIVA PPT.pptx
P.Surendar - VIVA PPT.pptx
 
Lung Conditions Prognosis Using CNN Model.pptx
Lung Conditions Prognosis Using CNN Model.pptxLung Conditions Prognosis Using CNN Model.pptx
Lung Conditions Prognosis Using CNN Model.pptx
 
BRAIN TUMOR DETECTION
BRAIN TUMOR DETECTIONBRAIN TUMOR DETECTION
BRAIN TUMOR DETECTION
 
computer aided detection of pulmonary nodules in ct scans
computer aided detection of pulmonary nodules in ct scanscomputer aided detection of pulmonary nodules in ct scans
computer aided detection of pulmonary nodules in ct scans
 
Journal article
Journal articleJournal article
Journal article
 
research on journaling
research on journalingresearch on journaling
research on journaling
 
CLASSIFICATION OF LUNGS IMAGES FOR DETECTING NODULES USING MACHINE LEARNING
CLASSIFICATION OF LUNGS IMAGES FOR DETECTING NODULES USING MACHINE LEARNING CLASSIFICATION OF LUNGS IMAGES FOR DETECTING NODULES USING MACHINE LEARNING
CLASSIFICATION OF LUNGS IMAGES FOR DETECTING NODULES USING MACHINE LEARNING
 
Classification of Lungs Images for Detecting Nodules using Machine Learning
Classification of Lungs Images for Detecting Nodules using Machine LearningClassification of Lungs Images for Detecting Nodules using Machine Learning
Classification of Lungs Images for Detecting Nodules using Machine Learning
 
Statistical Feature-based Neural Network Approach for the Detection of Lung C...
Statistical Feature-based Neural Network Approach for the Detection of Lung C...Statistical Feature-based Neural Network Approach for the Detection of Lung C...
Statistical Feature-based Neural Network Approach for the Detection of Lung C...
 
A REVIEW PAPER ON PULMONARY NODULE DETECTION
A REVIEW PAPER ON PULMONARY NODULE DETECTIONA REVIEW PAPER ON PULMONARY NODULE DETECTION
A REVIEW PAPER ON PULMONARY NODULE DETECTION
 
Automatic detection of tb
Automatic detection of tbAutomatic detection of tb
Automatic detection of tb
 
Multiple Analysis of Brain Tumor Detection Based on FCM
Multiple Analysis of Brain Tumor Detection Based on FCMMultiple Analysis of Brain Tumor Detection Based on FCM
Multiple Analysis of Brain Tumor Detection Based on FCM
 

Recently uploaded

一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
mzpolocfi
 
Criminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdfCriminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdf
Criminal IP
 
Adjusting OpenMP PageRank : SHORT REPORT / NOTES
Adjusting OpenMP PageRank : SHORT REPORT / NOTESAdjusting OpenMP PageRank : SHORT REPORT / NOTES
Adjusting OpenMP PageRank : SHORT REPORT / NOTES
Subhajit Sahu
 
Nanandann Nilekani's ppt On India's .pdf
Nanandann Nilekani's ppt On India's .pdfNanandann Nilekani's ppt On India's .pdf
Nanandann Nilekani's ppt On India's .pdf
eddie19851
 
一比一原版(Deakin毕业证书)迪肯大学毕业证如何办理
一比一原版(Deakin毕业证书)迪肯大学毕业证如何办理一比一原版(Deakin毕业证书)迪肯大学毕业证如何办理
一比一原版(Deakin毕业证书)迪肯大学毕业证如何办理
oz8q3jxlp
 
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
Timothy Spann
 
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
mbawufebxi
 
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
John Andrews
 
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
axoqas
 
Machine learning and optimization techniques for electrical drives.pptx
Machine learning and optimization techniques for electrical drives.pptxMachine learning and optimization techniques for electrical drives.pptx
Machine learning and optimization techniques for electrical drives.pptx
balafet
 
Adjusting primitives for graph : SHORT REPORT / NOTES
Adjusting primitives for graph : SHORT REPORT / NOTESAdjusting primitives for graph : SHORT REPORT / NOTES
Adjusting primitives for graph : SHORT REPORT / NOTES
Subhajit Sahu
 
一比一原版(BCU毕业证书)伯明翰城市大学毕业证如何办理
一比一原版(BCU毕业证书)伯明翰城市大学毕业证如何办理一比一原版(BCU毕业证书)伯明翰城市大学毕业证如何办理
一比一原版(BCU毕业证书)伯明翰城市大学毕业证如何办理
dwreak4tg
 
My burning issue is homelessness K.C.M.O.
My burning issue is homelessness K.C.M.O.My burning issue is homelessness K.C.M.O.
My burning issue is homelessness K.C.M.O.
rwarrenll
 
Malana- Gimlet Market Analysis (Portfolio 2)
Malana- Gimlet Market Analysis (Portfolio 2)Malana- Gimlet Market Analysis (Portfolio 2)
Malana- Gimlet Market Analysis (Portfolio 2)
TravisMalana
 
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
ahzuo
 
Unleashing the Power of Data_ Choosing a Trusted Analytics Platform.pdf
Unleashing the Power of Data_ Choosing a Trusted Analytics Platform.pdfUnleashing the Power of Data_ Choosing a Trusted Analytics Platform.pdf
Unleashing the Power of Data_ Choosing a Trusted Analytics Platform.pdf
Enterprise Wired
 
The Building Blocks of QuestDB, a Time Series Database
The Building Blocks of QuestDB, a Time Series DatabaseThe Building Blocks of QuestDB, a Time Series Database
The Building Blocks of QuestDB, a Time Series Database
javier ramirez
 
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
Timothy Spann
 
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptxData_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
AnirbanRoy608946
 
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Subhajit Sahu
 

Recently uploaded (20)

一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
 
Criminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdfCriminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdf
 
Adjusting OpenMP PageRank : SHORT REPORT / NOTES
Adjusting OpenMP PageRank : SHORT REPORT / NOTESAdjusting OpenMP PageRank : SHORT REPORT / NOTES
Adjusting OpenMP PageRank : SHORT REPORT / NOTES
 
Nanandann Nilekani's ppt On India's .pdf
Nanandann Nilekani's ppt On India's .pdfNanandann Nilekani's ppt On India's .pdf
Nanandann Nilekani's ppt On India's .pdf
 
一比一原版(Deakin毕业证书)迪肯大学毕业证如何办理
一比一原版(Deakin毕业证书)迪肯大学毕业证如何办理一比一原版(Deakin毕业证书)迪肯大学毕业证如何办理
一比一原版(Deakin毕业证书)迪肯大学毕业证如何办理
 
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
 
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
 
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
 
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
 
Machine learning and optimization techniques for electrical drives.pptx
Machine learning and optimization techniques for electrical drives.pptxMachine learning and optimization techniques for electrical drives.pptx
Machine learning and optimization techniques for electrical drives.pptx
 
Adjusting primitives for graph : SHORT REPORT / NOTES
Adjusting primitives for graph : SHORT REPORT / NOTESAdjusting primitives for graph : SHORT REPORT / NOTES
Adjusting primitives for graph : SHORT REPORT / NOTES
 
一比一原版(BCU毕业证书)伯明翰城市大学毕业证如何办理
一比一原版(BCU毕业证书)伯明翰城市大学毕业证如何办理一比一原版(BCU毕业证书)伯明翰城市大学毕业证如何办理
一比一原版(BCU毕业证书)伯明翰城市大学毕业证如何办理
 
My burning issue is homelessness K.C.M.O.
My burning issue is homelessness K.C.M.O.My burning issue is homelessness K.C.M.O.
My burning issue is homelessness K.C.M.O.
 
Malana- Gimlet Market Analysis (Portfolio 2)
Malana- Gimlet Market Analysis (Portfolio 2)Malana- Gimlet Market Analysis (Portfolio 2)
Malana- Gimlet Market Analysis (Portfolio 2)
 
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
 
Unleashing the Power of Data_ Choosing a Trusted Analytics Platform.pdf
Unleashing the Power of Data_ Choosing a Trusted Analytics Platform.pdfUnleashing the Power of Data_ Choosing a Trusted Analytics Platform.pdf
Unleashing the Power of Data_ Choosing a Trusted Analytics Platform.pdf
 
The Building Blocks of QuestDB, a Time Series Database
The Building Blocks of QuestDB, a Time Series DatabaseThe Building Blocks of QuestDB, a Time Series Database
The Building Blocks of QuestDB, a Time Series Database
 
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
 
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptxData_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
 
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
 

Lung nodule diagnosis from CT images based on ensemble learning

  • 1. Lung Nodule Diagnosis from CT Images Based on Ensemble Learning Mohammad Hossein Fazel Zarandi Farzad Vasheghani Farahani Abbas Ahmadi 2015 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology Niagara Falls, Canada | August 12 - 15
  • 2. Topics • Objective • Procedure • Background • Proposed system • Contribution • Conclusion • References
  • 3. Objective The purpose of this study is to “design a CAD system for diagnosis of lung cancer” which is: • Helping physicians as a second opinion • Fast and accurate for early diagnosis of this disease • Based on processing CT images and image analysis techniques • Using ensemble of classifiers
  • 4. Procedure How to achieve the desired objectives?
  • 6. About lung cancer • Cancer is one of top mortal diseases in the world [1] like heart diseases and diabetes. • Lung cancer is the most leading cause of deaths in both women and men (27.2% of all cancer deaths) [2]. • Lung cancer is caused by an uncontrollable irregular growth of cells in lung tissue. • Lung tissue abnormalities that are roughly spherical with round opacity and a diameter of up to 30mm [3] are known as lung nodules.
  • 7. Computer-aided diagnosis (CAD) systems • CAD systems Are very helpful for radiologists in the detection and diagnosing abnormalities earlier and faster [4]. • Indeed, computer-aided diagnosis systems act as a second opinion for radiologists before suggesting a biopsy test [5].
  • 8. Machine learning methods • Ensemble learners combine decisions of base classifiers to form an integrated output [8]. • Machine learning methods have become one of the dominant approaches in CAD for medical imaging [6]. • One of the four pioneer branches in the field of machine learning is ensemble based systems [7].
  • 11. Lung CT dataset • 60 CT images including both men and women collected by Lung Image Database Consortium (LIDC). • Lung lesions in these images are marked by up to four radiologists in VIA group. • All the lung CT images are in the format of DICOM with the size of 512*512.
  • 12. Image processing phase Image processing module in this model consists of 4 main steps: Preprocessing Segmentation Post processing Feature extraction Raw images Extracted features
  • 13. 1) Preprocessing • Image preprocessing is a process which eliminates primary noise and image distortion, and also enhances important features exists in CT images. • By using some image enhancement methods including median filter and histogram equalization.
  • 14. 2) Segmentation a) Background elimination and other additional parts such as bones to extract lung tissue and region of interest (ROI) in the lung image. b) Convert the image to black and white in order to prepare it for boundary tracing. Method: Region growing Method: Thresholding
  • 15. 2) Segmentation (Cont.) Background elimination using region growing Convert to black and white using thresholding
  • 16. 3) Post processing • Eliminate unwanted objects (small or large) inside and outside the binary image of lungs. • Join the detected object in the image together by filling in the gaps between them and by smoothing their outer edges. • Fill image regions and holes of the detected objects. • Find boundaries of the objects by concentrating only on exterior boundaries.
  • 17. 4) Feature extraction In order to classify suspected objects as nodule or non-nodule, further working is needed to extract specific features from raw images. Feature Definition Roundness 4𝐴 𝜋𝐿2 Circularity 4𝜋𝐴 𝑃2 Compactness 𝑃2 4𝜋𝐴 Ellipticity 𝜋𝐿2 2𝐴 Eccentricity − • In this study, five morphological features are extracted for each suspected objects from CT images. • P, A and L denote the perimeter, area and maximum diameter of the suspected objects, respectively.
  • 18. Classification phase One of the suitable methods to improve the accuracy of classification is the use of multiple classifier systems. Multiple classifier systems generally consist of two main parts [9]: • Creation an ensemble • Combination of class labels
  • 19. 1) Creating an ensemble Three different classifiers including Multilayer perceptron (MLP), K-Nearest neighbor (KNN), and Support vector machines (SVM) work on numerical data obtained from feature vectors. Note: The extracted features from suspected objects are considered as inputs to the ensemble system and they are normalized between zero and one. Hence, the feature set consists of five extracted features from the previous section and a binary target feature that defines suspected objects are nodule or not (0 or 1).
  • 20. 2) Combining outputs Each base classifier has its own opinion and identifies suspected objects as nodule or non-nodule. Majority voting techniques is used to combine the results of base classifiers. Final output of ensemble system
  • 21. System Performance Measurements Accuracy (Acc), Sensitivity (Sn) , Specificity (Sp), G-mean , Precision (Prc) and F-measure of the proposed system: Proposed method System Performance Measurements Acc Sn Sp G-mean Prc F-measure MLP 90.41 73.55 94.68 83.45 77.76 75.60 KNN 91.20 81.76 93.59 87.48 76.33 78.96 SVM 90.60 73.44 94.94 83.50 78.59 75.93 Majority Voting 94.65 84.16 97.30 90.49 88.76 86.40
  • 22. Confusion matrix MLP Actual Predicted Negative Positive Negative 75.56 % 4.25 % Positive 5.34 % 14.85 % KNN Actual Predicted Negative Positive Negative 74.69 % 5.12 % Positive 3.68 % 16.51 % SVM Actual Predicted Negative Positive Negative 75.77 % 4.04 % Positive 5.36 % 14.83 % Proposed method Actual Predicted Negative Positive Negative 77.66 % 2.15 % Positive 3.20 % 16.99 %
  • 23. Contribution The main contribution of this paper is to: present a classifier ensemble system that uses image processing and data mining techniques for lung nodule identification. • Feature extraction of suspected objects in CT images is the main goal of image processing phase. • In data mining phase, there are three different classifiers including MLP, KNN and SVM that work on extracted features to identify lung objects as nodule or non-nodule in the form of group decision- making.
  • 24. Conclusion • In present work, the manner of a CAD system in diagnosis process is simulated. • Our proposed CAD system is a highly promising method, providing high performance in the identification of pulmonary nodules from CT images. • For the sake of checking the effectiveness of the proposed hybrid intelligent method, 5-fold cross-validation approach was applied and the program ran for 10 times. • Performance measurements show that the system could be used as a second opinion for physicians in the diagnosis of lung cancer from CT images.
  • 25. References 1. “WHO | The top 10 causes of death.” 2. “Cancer of the Lung and Bronchus - SEER Stat Fact Sheets.” [Online]. Available: http://seer.cancer.gov/statfacts/html/lungb.html. [Accessed: 13-Feb-2015]. 3. J. H. M. Austin, P. Naidich, L. Muller, M. Hansell, and A. Zerhouni, “Glossary of terms for CT of the lungs: recommendations of the Nomenclature Committee of the Fleischner Society,” Radiology, vol. 200, no. 2, pp. 327–331, 1996. 4. R. N. Strickland, “Tumor detection in non stationary backgrounds,” IEEE Trans. Med. Imaging, vol. 13, no. 3, pp. 491–9, Jan. 1994. 5. S. B. Lo, S. A. Lou, J. Lin, M. T. Freedman, M. V Chien, and S. K. Mun, “Artificial Convolution Neural Network Techniques and Applications for Lung Nodule Detection,” Med. Imaging, IEEE Trans., vol. 14, no. 4, pp. 711–718, 1995. 6. S. L. a. Lee, a. Z. Kouzani, and E. J. Hu, “Automated identification of lung nodules,” 2008 IEEE 10th Work. Multimed. Signal Process., pp. 497–502, Oct. 2008. 7. T. G. Dietterich, “Machine-Learning Research,” AI Mag., vol. 18, no. 4, pp. 97–136, 1997. 8. S. L. a Lee, a Z. Kouzani, and E. J. Hu, “Random forest based lung nodule classification aided by clustering.,” Comput. Med. Imaging Graph., vol. 34, no. 7, pp. 535–42, Oct. 2010. 9. R. Polikar, “Ensemble Based Systems in Decision Making,” Circuits Syst. Mag. IEEE, vol. 6, no. 3, pp. 21–45, 2006.

Editor's Notes

  1. 1