computer methods and programs in biomedicine 95 ( 2 0 0 9 ) 270–279 journal homepage: www.intl.elsevierhealth.com/journals/cmpb Discrimination of myocardial infarction stages by subjective feature extraction Dingfei Ge a,* , Lihui Sun a , Jiayin Zhou b , Yuquan Shao c a School of Information and Electronic Engineering, Zhejiang University of Science and Technology, Hangzhou 310023, China b Biomedical Engineering Research Centre, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore c Sir Run Run Shaw Hospital, Affiliated Hospital of Medical College, Zhejiang University, Hangzhou 310016, China article info Article history: Received 12 July 2008 Received in revised form 6 December 2008 Accepted 22 March 2009 Keywords: Myocardial infarction ECG Feature extraction Subjective-classification Hyper-dimensional time-series abstract Lots of studies on myocardial infarction (MI) computer assisted diagnosis are based on cer- tain important ECG components which only account for local information. 12-Lead ECG signals which were regarded as hyper-dimensional time-series data were utilized to extract features from global information in this study. Existing feature extraction techniques for classification attempt to classify all the classes included. However sometimes it is more important to better recognize certain specific classes rather than to discriminate all the classes. A feature extraction method based on subjective-classification was proposed to discriminate the specific classes, which the classification priority was given subjectively, and each of the other classes was separated at the same time. The method includes data reduction by principal component analysis (PCA), data normalization by whitening transfor- mation and derivation of projecting vectors for subjective-classification, etc. The data in the analysis were collected from PTB diagnostic ECG database. The results show that the pro- posed method can obtain a small number of effective features from 12-lead ECGs to better classify classes with priority, and the other classes can be classified at the same time. © 2009 Elsevier Ireland Ltd. All rights reserved. 1. Introduction Automatic detection and classification of abnormalities in electrocardiogram (ECG) will be of great help in medical exam- inations or monitoring of critical ill patients. Lots of existing studies on myocardial infarction (MI) computer assisted diag- nosis are based on certain important ECG components which only account for local information, such as VCG and ECG based MI detection [1], Neural Network aided MI diagno- sis [2], time-domain threshold methods for MI classification [3], detecting MI using ECG parameters [4], Hermite expan- sions [5], multiple logistic regression method [6]. Actually, clinical physicians diagnose patient’s diseases are usually based on 12-lead ECG and entire ECG cycle. In order to ∗ Corresponding author. Tel.: +86 13186950314. E-mail address: gedingfei@hotmail.com (D. Ge). obtain global information, 12-lead ECG and its entire ECG cycle were used for the analysis in current study. The key problem of the classification is how to obtain essential information from the hyper-dimensional data. Feature selec- tion and feature extraction are two major approaches to dimensionality reduction. Feature selection seems to lose some of significant information for the classification due to selecting a subset of the original input variables in the measurement space, while feature extraction involves a transformation of the original variables and the features pro- vided are a set of new variables in the transformed space. Generally, feature extraction provides more efficient represen- tation of patterns [7]. Thus, feature extraction technique for dimensionality reduction is adopted in this study. The ECG 0169-2607/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.cmpb.2009.03.008