Intelligent methods for capnogram feature extraction applied to asthma classification
Betancourt Pomares, Janet Pomares · Institutional Repositories DataBase (IRDB) · 2014
A method for ECG and capnogram signals classification is proposed based on fuzzy similarity evaluation, where shape exchange algorithm and fuzzy inference are combined.It aims to be applied to quasi-periodic biomedical signals and has low computational cost.On the experiments for atrial fibrillation (AF) classification using two databases: MIT-BIH AF and MITBIH Normal Sinus Rhythm, values of 100%, 94.4%, and 97.6% for sensitivity, specificity, and accuracy respectively, and execution time of 0.6 s are obtained.As for the capnogram database, experiments for asthma classification produce 83.3%, 80.9%, and 81.5 for sensitivity, specificity, and accuracy respectively, and execution time of 0.3 s.The proposal is capable of been extended to classify different diseases, from ECG and capnogram signals, such as: Brugada syndrome, AV block, hypoventilation, and asthma among others to be implemented in low computational resources devices.To improve the previous classification results, a method for feature extraction of the capnogram based on wavelet decomposition for asthma classification is proposed.The method requires low computational cost and shows adequate performance for a real time classification of asthma severity.The validation experiments on 23 capnograms, collected from an Asthma Camp in Cuba, suggest 97.39% of accuracy as best classification result using a SVM classifier.An estimation of the execution time for a II physiological multiparameter monitor is obtained, showing an average of 8 sec to determine the suitable features.The proposal aims to be a part in the decision support system for asthma classification that is under development by research group of Tokyo Institute of Technology (TITECH) and Tokyo Dental and Medical University (TMDU).To integrate the two proposals in this investigation and evaluate their influence in the classification result, two different architectures for the decision support system are evaluated.Both architectures show similar classification results in terms of correct rate.Execution time represents the most significant difference between architectures and could define the kind of application where to be employed.