Methodenanalyse und -optimierung der Impedanzplethysmographie zur Diagnostik kardialer und arterieller Erkrankungen
Abstract The objectives of this thesis are the determination and optimization of clinical applicability of impedance plethysmography for the diagnostics of cardial and arterial diseases. It is shown that the impedance plethysmographical determination of physiological parameters is controversial. In opposition to that, monitoring of therapeutic actions is acknowledged to a large extent. The analysis of literature reveals only qualitative results regarding diagnostics of cardiovascular diseases by means of impedance cardiography. The peripheral application of diagnosing peripheral arterial occlusive disease (PAOD) by means of impedance plethysmography has achieved good quantitative results. In order to quantify the diagnostic potential for early detection of cardiovascular diseases a clinical study has been undertaken in the department of thoracic, heart and vascular surgery at Fulda hospital. 118 thoracic and peripheral impedance measurements were performed at 85 patients. Moreover, 18 control subjects were included. The results of this theses show that the early detection of mitral insufficiency by means of impedance cardiography can be achieved. The classification procedures yield 82% of sensitivity (Se) and 94% of specificity (Sp). Furthermore, the usage for diagnosing coronary heart disease is recommended (Se=95%, Sp=94%). In contrary to literature, detection of aortic valve diseases seems not feasible (Se<35%, Sp=80…90%). The early diagnostics of PAOD by means of impedance plethysmography should be promoted (Se=100%, Sp=93%) especially due to the introduction of a method which enables the examination independent of measurement conditions by means of analysis of peripheral state curve. Moreover, the new method Similarity-Averaging is described that enlarges the field of application of impedance cardiography to patients suffering from continuous arrhythmia. Besides, the selection of a method for transformation to the time-frequency domain is achieved in an unbiased manner even if the coefficient distributions are qualitative different.