Morphological Postprocessing for Color Flow Imaging in Medical Ultrasound

Color flow imaging in medical ultrasound color codes pixels in blood vessels according to its estimated velocity, which is calculated at every pixel within a user selected region of interest regardless of whether the pixel came from tissue or blood flow. Hence, a tissue/flow detector is required to color code only flow pixels. Scanning conditions and scenarios, available output parameters from the chosen estimator, and system noise or other sources of error may pose fundamental difficulties in making the correct tissue/flow decision at all spatial locations and all temporal frames. Therefore, post-processing on the decision data may be necessary for robust detection. We propose a morphological spatial method based on complementary hulling that significantly improves the accuracy of tissue/flow detection while improving the accuracy of velocity estimates. We compare the method to other spatial techniques from in-vivo and in-vitro data. Our method is shown to be more accurate at removing both false positive and false negative decisions and corrects noisy velocity estimations.