Site Map
To this day, the image quality of ultrasonic images is still suffers from poor resolution and speckle. An array of transducers and beamforming is almost always used for improved image quality. In this article, image characteristics, delay-and-sum beamforming and minimum variance beamforming will be discussed.
The two characteristics of ultrasound imaging are resolution and speckle. Resolution can be defined as the minimum distance between which two point scatters need to be from each in order for them to be image as individual sources. Poor resolutions results in being unable to decipher which waves are coming from which point scatters.
Another characteristic of ultrasonic images is speckle. Speckle gives ultrasonic images their grainy look. This artifact results when a large number of densely distributed point scatters that are too small be resolved by the imaging system reflect a transmitted signal. The combination of all the reflections results in a signal with a Rayleigh distributed magnitude. This causes the spatial correlation of the resulting image to differ from the actual. Speckle regions can either be considered noise or useful information. This depends on the ultrasonic imaging application. For example, in medical imaging, speckle may be an important characteristic because they give an indication of homogenous tissues.
In ultrasonic imaging a large array of transducers are used for imaging. With the added spatial dimension beamforming can be applied to the imaging process to improve image quality. Usually, delay-and-sum beamforming (DAS) or minimum variance (MV) beamforming are used to scan the image so only reflections directly from the point scatters are imaging and the multi-path reflections and reflections from other point scatters not within beampattern are filtered out. In delay-and-sum beamforming the beamformer vectors (the weights applied to received data) are predetermined to produce a narrow mainlobe while minimizing the sidelobes. For MV beamforming the beamforming vectors are based on the statistics of the received data. The object of a MV beamformer is to minimize the variance of the output of the beamformer, while maintaining unit gain in the desired direction. By using the statistics (a covariance matrix) of the recorded the signal beamformer adaptively adjusts its beampattern to steer nulls in the direction of interferes (reflections from scatters not from the desired direction), so the output beamformer has little deviation.
The MV beamformer can achieve better resolution than the DAS beamformer because of its adaptive nature. But it is to be noted that because the weights of the beamforming vectors are calculated from the statistics of the received data, a MV beamformer will affect speckle characteristics differently than a DAS. Generally, MV reduces the magnitude of speckle. If speckle information is important in the medical application, then the MV beamformer has to be adjusted. By using spatial and temporal averging of the covariance matrix, one can control the resolution and the magnitude of speckle. Increasing the length of the subarray, results in spatial averaging which improves the resolution of the image because it improves the estimate of the covariance matrix. This also results in a reduction of speckle magnitude. For a given subbarray length, increasing the length of the temporal average for the MV beamformer increases the intensity of the speckle with the resolution being maintained.