Estimating frequency dependent attenuation to improve autmatic time gain compensation in B-mode imaging

The last decade, there has been a trend towards smaller ultrasound equipment. For the pocket size scanners, there is no space for all the buttons and sliders found on full-size scanners. Automatic setting of gain is of utmost importance. This work presents a method for estimating frequency dependent attenuation as a step in improved setting of time gain compensation (TGC). Most TGC algorithms work with envelope detected data. Thus, they may introduce artifacts and hide useful clinical information. This method attacks the problem from a signal processing view. It is based on linear acoustics and signal processing techniques applied to radio frequency (RF) data. First, a frequency estimator is applied to the received RF data. The result is used for estimation of the frequency dependent attenuation which, in turn, is used in the calculation of a TGC curve. The method is verified by simulation (Field II [1] [2]) and phantom experiments. The transducer is a phased linear array transducer (GE Vingmed M3S) using a broadband transmit pulse with a center frequency of 2.5MHz. The phantom is a CIRS40 (CIRSINC) tissue mimicking phantom. The frequency downshift with depth is found by short time fourier transform (STFT) and autoregressive (AR) spectrum estimators of first and second order on qudrature demodulated and RF data respectively. The diffraction effect is compensated by a modified method earlier used in tissue characterisation [3]. The scattering from tissue is assumed to behave as isotropic Rayleigh scattering. Based on experiments we show how it is possible, in simple cases, to calculate the power loss due to frequency dependent attenuation depending on the scanning object. It constitutes an alternative way of applying tissue characterisation methods to improve "Auto focus" features on modern ultrasound scanners.