Computer-Aided Lung Nodule Recognition by SVM Classifier Based on Combination of Random Undersampling and SMOTE
Authors:
Open Access
- 1 January 2015
- journal article
- research article
- Published by Wiley in Computational and Mathematical Methods in Medicine
- Vol. 2015, 1-13
Abstract
In lung cancer computer-aided detection/diagnosis (CAD) systems, classification of regions of interest (ROI) is often used to detect/diagnose lung nodule accurately. However, problems of unbalanced datasets often have detrimental effects on the performance of classification. In this paper, both minority and majority classes are resampled to increase the generalization ability. We propose a novel SVM classifier combined with random undersampling (RU) and SMOTE for lung nodule recognition. The combinations of the two resampling methods not only achieve a balanced training samples but also remove noise and duplicate information in the training sample and retain useful information to improve the effective data utilization, hence improving performance of SVM algorithm for pulmonary nodules classification under the unbalanced data. Eight features including 2D and 3D features are extracted for training and classification. Experimental results show that for different sizes of training datasets our RU-SMOTE-SVM classifier gets the highest classification accuracy among the four kinds of classifiers, and the average classification accuracy is more than 92.94%.Keywords
Funding Information
- National Natural Science Foundation of China (61370152, N130204003, 201202060, 2014BAI17B02)
- Ministry of Higher Education through Fundamental Research Grant Scheme for the Central Universities (61370152, N130204003, 201202060, 2014BAI17B02)
- Natural Science Foundation of Liaoning Province (61370152, N130204003, 201202060, 2014BAI17B02)
- Ministry of Science and Technology of the People’s Republic of China (61370152, N130204003, 201202060, 2014BAI17B02)
This publication has 30 references indexed in Scilit:
- Potential Lung Nodules Identification for Characterization by Variable Multistep Threshold and Shape Indices from CT ImagesComputational and Mathematical Methods in Medicine, 2014
- Pulmonary nodule registration in serial CT scans using global rib matching and nodule template matchingComputers in Biology and Medicine, 2014
- Random forest based lung nodule classification aided by clusteringComputerized Medical Imaging and Graphics, 2010
- Adaptive border marching algorithm: Automatic lung segmentation on chest CT imagesComputerized Medical Imaging and Graphics, 2008
- Automated detection of lung nodules in CT images using shape-based genetic algorithmComputerized Medical Imaging and Graphics, 2007
- Lung metastases detection in CT images using 3D template matchingMedical Physics, 2007
- Autonomous detection of pulmonary nodules on CT images with a neural network-based fuzzy systemComputerized Medical Imaging and Graphics, 2005
- Pulmonary Nodules on Multi–Detector Row CT Scans: Performance Comparison of Radiologists and Computer-aided DetectionRadiology, 2005
- Three-dimensional segmentation and growth-rate estimation of small pulmonary nodules in helical ct imagesIEEE Transactions on Medical Imaging, 2003
- Computer-aided diagnosis in chest radiography: a surveyIEEE Transactions on Medical Imaging, 2001