Automatic Lung Nodule Detection Using Improved MKM Clustering Algorithm (IMKM) and Gentle Boost Classifier in CT Images

Lung Computer-Aided Diagnosis (CAD) method provided precise analysis of Computer Tomography (CT) images. This paper introduces a progressed lung nodule detection method with high precision. Different steps in diagnosing lung nodules include pre-processing, segmentation, candidate extraction and nodu...

Full description

Saved in:
Bibliographic Details
Published inInternational journal of computer science and information security Vol. 14; no. 11; p. 764
Main Authors Jafari, Mitra, Fazli, Saeid
Format Journal Article
LanguageEnglish
Published Pittsburgh L J S Publishing 01.11.2016
Subjects
Automation
Classification
Clustering
Datasets
Genetic algorithms
Lung cancer
Methods
Neural networks
Online AccessGet full text

Cover

More Information
Summary:Lung Computer-Aided Diagnosis (CAD) method provided precise analysis of Computer Tomography (CT) images. This paper introduces a progressed lung nodule detection method with high precision. Different steps in diagnosing lung nodules include pre-processing, segmentation, candidate extraction and nodule detection. In pre-processing step lungs extracted from the chest region. In segmentation step we proposed an improved moving k-means clustering algorithm (IMKM) which resolve problems in previous algorithms. For classification we faced with the problem of imbalanced data. The majority of samples are belonging to the class of non-nodule organs and the minority of them are nodules. In this paper we applied Gentle Boost method which is a cost sensitive algorithms for classification. First of all, some features extracted from segmented areas to train classification algorithm. Then we apply these features to the algorithm. Simulation results shows that the proposed algorithm can detect nodules with high performance of AUC=90% and reduce false positive up to 3.8%.
ISSN:1947-5500