Résumé:
The work realized in this thesis presents processing and analysis techniques of the
electrocardiographic signal (ECG) as well as the respiratory signal for the classification and
discrimination of cardiac and respiratory diseases based on fractional order operators and
systems. First, the frequency content of the complex QRS of the ECG modeling by a
fractional order linear system was presented. After that, classification and discrimination
methods of cardiac arrhythmias have been developed using the proposed complex QRS
frequency content fractional order model’s parameters as pertinent parameters. A particular
attention is paid to normal beats, premature ectopic (PVC) and right and left branch blocks
(RBBB and LBBB). Some classification experiments have been presented using the MIT /
BIH database. Then, the respiratory impedance modeling and the respiratory diseases
classification using fractional order operators and systems have been also presented. The
respiratory impedance model parameters have been used by a classification algorithm for the
healthy subjects, asthmatic and chronic obstructive pulmonary disease (COPD) patient’s
discrimination. The classification method was validated using data collected by the forced
oscillation technique (FOT) at Ghent University Hospital, Belgium. The obtained results for
the frequency content of the QRS and the respiratory impedance modeling and the cardiac and
respiratory diseases discrimination were very satisfactory.
