Pairwise FCM based feature weighting for improved classification of vertebral column disorders.
In this paper, an innovative data pre-processing method to improve the classification performance and to determine automatically the vertebral column disorders including disk hernia (DH), spondylolisthesis (SL) and normal (NO) groups has been proposed. In the classification of vertebral column disor...
| Publicado en: | Computers in Biology & Medicine Vol. 46; pp. 61 - 71 |
|---|---|
| Autores principales: | , , |
| Formato: | research Journal Article |
| Publicado: |
Elsevier B.V.
2014
|
| Acceso en línea: | Ver este registro en EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=ccm&AN=104026612&site=ehost-live header: @attributes: shortDbName: ccm uiTerm: 104026612 longDbName: CINAHL Complete uiTag: AN controlInfo: bkinfo: dissinfo: jinfo: jid: 00104825 JC2 jtl: Computers in Biology & Medicine issn: 00104825 maglogo: N pubinfo: dt: 2014 vid: 46 pid: 82545 pub: Elsevier B.V. place: Philadelphia, Pennsylvania artinfo: ui: 104026612 NLM24529206 2012480763 10.1016/j.compbiomed.2013.12.004 NLM24529206 104026612 ppf: 61 ppct: 10 formats: tig: atl: Pairwise FCM based feature weighting for improved classification of vertebral column disorders. aug: au: Unal, Yavuz Polat, Kemal Erdinc Kocer, H affil: Education Faculty, Amasya University, Amasya, Turkey. sug: subj: Magnetic Resonance Imaging Models, Biological Spinal Diseases Classification Spinal Diseases Radiography Spine Radiography Algorithms Resource Databases ab: In this paper, an innovative data pre-processing method to improve the classification performance and to determine automatically the vertebral column disorders including disk hernia (DH), spondylolisthesis (SL) and normal (NO) groups has been proposed. In the classification of vertebral column disorders' dataset with three classes, a pairwise fuzzy C-means (FCM) based feature weighting method has been proposed. In this method, first of all, the vertebral column dataset has been grouped as pairwise (DH-SL, DH-NO, and SL-NO) and then these pairwise groups have been weighted using a FCM based feature set. These weighted groups have been classified using classifier algorithms including multilayer perceptron (MLP), k-nearest neighbor (k-NN), Naive Bayes, and support vector machine (SVM). The general classification performance has been obtained by averaging of classification accuracies obtained from pairwise classifier algorithms. To evaluate the performance of the proposed method, the classification accuracy, sensitivity, specificity, ROC curves, and f-measure have been used. Without the proposed feature weighting, the obtained f-measure values were 0.7738 for MLP classifier, 0.7021 for k-NN, 0.7263 for Naive Bayes, and 0.7298 for SVM classifier algorithms in the classification of vertebral column disorders' dataset with three classes. With the pairwise fuzzy C-means based feature weighting method, the obtained f-measure values were 0.9509 for MLP, 0.9313 for k-NN, 0.9603 for Naive Bayes, and 0.9468 for SVM classifier algorithms. The experimental results demonstrated that the proposed pairwise fuzzy C-means based feature weighting method is robust and effective in the classification of vertebral column disorders' dataset. In the future, this method could be used confidently for medical datasets with more classes. pubtype: Academic Journal doctype: research Journal Article ougenre: Article language: English refInfo: holdings: @attributes: islocal: N |
|---|