Joint probability discrimination between stationary tissue and blood velocity signals

In CFM-mode the blood velocity estimates are overlaid onto the B-mode image. The velocity estimation gives non-zero velocity estimates in both the surrounding tissue and the vessels. A discrimination algorithm is needed to determine, which estimates represent blood flow and should be displayed. This study presents a new statistical discriminator. Investigation of the RF-signals reveals that features can be derived that distinguish the segments of the signal, which do an do not carry information on the blood flow. In this study 4 features, have been determined: (a) the energy content in the segments before and after echo-canceling, and (b) the amplitude variations between samples in consecutive RF-signals before and after echo-canceling. The statistical discriminator was obtained by computing the probability density functions (PDFs) for each feature through histogram analysis of data. The discrimination is performed by determining the joint probability of the features for the segment under investigation and choosing the segment type that is most likely. The method was tested on simulated data resembling RF-signals from the carotid artery.