Mathematische Analyse des Dopplersignals zur quantitativen Bestimmung des Blutflusses.

Modern radiologic imaging aims to give quantitative parameters, e. g. blood circulation, as well as morphological representation. But until now Doppler sonography could only measure blood flow velocities: The received Doppler signals are processed using the Fast Fourier Transformation (FFT). The amplitudes of the resulting velocity spectra are either encoded as brightnesses and plotted as a function of time and frequency shift to provide a two-dimensional spectral display or the average Doppler shift is encoded as a color and superimposed to the B-mode image (Color Doppler). Both methods only allowed a semi quantitative conclusion on the actual blood flow. Can a statement be made through the use of special analysis of the blood flow, by providing the examiner with non-invasive quantitative information about the blood circulation in tissue, in contrast to semi-quantitative parameters of the ultra sound system? Since the signals are mainly backscattered by the red blood cells, this work wants to examine how far the measured intensity is dependent on the blood flow or on the hematocrit. By knowing the correlation it should be possible to calculate in reverse the gross flow by measuring the intensity and the hematocrit. This should also be possible with different flow profiles. First a flow model was developed, which was able to generate physiological and pathophysiological flow profiles with independently adjustable gross flows and flow profiles. A catheter (calibre 0.7 mm) was positioned in a basin of outgased Butanediol. Butanediol has a similar impedance to human tissue. Blood in different diluted states was used as a flow medium, as well as the contrast agent SonoVue ®. A steady flow was created by a perfusion pump behind which a microprocessor controlled flow modulator was used. With this flow model it was possible to produce reliable and reproducible flow profiles. The Doppler ultrasound signals were saved as raw rf-data, and were transferred directly behind the beam former onto an external computer via the use of the Ultrasound Research Interface®. Data analysis happened only within the frequency domain and was programmed with MATLAB®. The rf-data was evaluated with the URI-OPT-Package® for MATLAB developed by UC Davis. The included algorithms were considerably modified and further improved. The rf-signal was off-line processed into its frequency spectrum on a PC using the fast Fourier transformation and displayed 3-dimensionally. The Doppler spectrum was integrated through its frequency field as well as over time. This enabled the calculation of the entire intensity of all Doppler frequencies per time. Finally, the correlation significance of the intensity to the hematocrit and the flow was assessed. Air bubbles could falsify the measured result through their high echogenicity and since it was not entirely possible to eliminate all of them further algorithms were established to mathematically eliminate their backscattered signals. For the measured entire intensity, the dependency on the hematocrit and on the pre-installed gross flow was calculated. It showed a significant correlation with the pre-installed gross flow, but because of the rouleaux formation of the red blood cells the intensity was not linearly correlated to the hematocrit. It was shown that the intensity of the Doppler signal is maximized at a hematocrit of about 0.15. The result could not be transferred from diluted series to physiological situations, since the hematocrit of native blood lies between 0.3 and 0.5. When blood is diluted the rouleaux formation breaks up, therefore the regression lines of the intensity of native blood is distinctively flatter. It was also shown that the flow profile, and accordingly the turbulence in the suspension, has a significant impact on the measured intensity of the Doppler signal. If, however, the flow profile and hematocrit are known it was possible to categorize the gross flow volume due to empirical values. Additionally, this work showed an affordable and reliable flow model, which enabled to cut out different effects. Because of this it can be used for various different problems. Provided the flow profile and hematocrit are known, it was possible to use the here shown correlations to ascertain the gross flow by its measured intensity. From these results it was concluded it is possible to determine the quantity of blood circulation via ultrasound on the basis of the here developed algorithms. Constants for calculation were derived and the shown measured results of the model can be now used in clinical work situations.