|
Accurate measurement of tissue aberrations is necessary for effective adaptive ultrasound imaging. Higher order arrays provide more elements and a larger array footprint over which echo signals can be acquired. This allows for better sampling of the aberrator in both the azimuthal and elevation dimensions. These measured aberration profiles can then be used to correct the timing of transmitted and received RF signals to generate new images. We acquired single channel RF data on a 6.7 MHz, 8 x 128 array (Tetrad Co.) operating at F/1.0 in azimuth and F/2.9 in elevation. This array was interfaced to a Siemens Elegra scanner, allowing for data acquisition during routine phantom and clinical scanning. One-dimensional and two-dimensional physical near-field aberrators were used while imaging speckle only and spherical cyst-mimicking phantoms. In some experiments, neighboring elements were electronically tied in elevation to form ``taller'' elements. We collected individual channel data on each of 6 physical rows and then on a combination of rows to form 3x128, 2x128, and 1x128 arrays over a 6x128 aperture of the array. A least-mean-squares algorithm was employed to estimate the arrival time error induced by the tissue for the different array geometries. These aberration measurements were used to correct the images. In addition, point target simulations were performed to characterize the algorithm's performance for all four different array configurations. We present the performance of the adaptive imaging algorithm and discuss methods of combining arrival time profiles from axial and lateral tissue regions to improve adaptive imaging performance. Contrast results in simulation and phantom experiments with different aberrators are presented. We also discuss, in the context of our aberration measurement profiles, the array geometry requirements for successful adaptive imaging and the effects of the aberrators on sidelobe strength and contrast measurement. Results from performing adaptive imaging on clinical breast images using a 6x128 array geometry are also presented.
|
