心臓の超音波断層像における心臓壁領域の自動同定に関する検討

抄録

For the assessment of cardiac functions using ultrasonography, several researchers have reported the efficiency of the two-dimensional myocardial displacement measurement. However, it requires the identification of heart wall areas. It is important to identify heart wall areas automatically for the convenience and the elimination of the examiner dependence. Previously, the heart wall areas are estimated from the echo intensity of B-mode images, and it is difficult to identify the heart wall areas with low echo intensity. In the present study, we propose a novel feature, magnitude squared coherence function (MSC), which is effective for automatic heart wall identification. This feature represents the time changes of the ultrasonic RF signals. We have identified heart wall areas using a linear discriminant analysis with MSC. Most of the identified areas agreed with those with high echo intensity, showing a great potential of the proposed feature for heart wall identification.