Lung nodule segmentation and recognition using SVM classifier and active contour modeling: A complete intelligent system

Authors:
Mohsen Keshani;Zohreh Azimifar;Farshad Tajeripour;Reza Boostani
Affiliations:
Department of Computer Science and Engineering, School of Electrical and Computer Engineering, Shiraz University, Shiraz, Iran;Department of Computer Science and Engineering, School of Electrical and Computer Engineering, Shiraz University, Shiraz, Iran;Department of Computer Science and Engineering, School of Electrical and Computer Engineering, Shiraz University, Shiraz, Iran;Department of Computer Science and Engineering, School of Electrical and Computer Engineering, Shiraz University, Shiraz, Iran
Venue:
Computers in Biology and Medicine
Year:
2013

Citing 4
Cited 0

A new efficient SVM-based edge detection method

Pattern Recognition Letters
Detection and Recognition of Lung Abnormalities Using Deformable Templates

ICPR '04 Proceedings of the Pattern Recognition, 17th International Conference on (ICPR'04) Volume 3 - Volume 03
Blob Segmentation Using Joint Space-Intensity Likelihood Ratio Test: Application to 3D Tumor Segmentation

CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 2 - Volume 02
Localizing Region-Based Active Contours

IEEE Transactions on Image Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, a novel method for lung nodule detection, segmentation and recognition using computed tomography (CT) images is presented. Our contribution consists of several steps. First, the lung area is segmented by active contour modeling followed by some masking techniques to transfer non-isolated nodules into isolated ones. Then, nodules are detected by the support vector machine (SVM) classifier using efficient 2D stochastic and 3D anatomical features. Contours of detected nodules are then extracted by active contour modeling. In this step all solid and cavitary nodules are accurately segmented. Finally, lung tissues are classified into four classes: namely lung wall, parenchyma, bronchioles and nodules. This classification helps us to distinguish a nodule connected to the lung wall and/or bronchioles (attached nodule) from the one covered by parenchyma (solitary nodule). At the end, performance of our proposed method is examined and compared with other efficient methods through experiments using clinical CT images and two groups of public datasets from Lung Image Database Consortium (LIDC) and ANODE09. Solid, non-solid and cavitary nodules are detected with an overall detection rate of 89%; the number of false positive is 7.3/scan and the location of all detected nodules are recognized correctly.