An in vitro study on porcine skin: attenuation profile estimation using auto-regressive modeling

The frequency-dependent attenuation is an interesting acoustic parameter for tissue characterization. For in vivo skin characterization, the anatomy of this biological tissue requires an echographic technique since transmission techniques are not available. In case of highly attenuating medium such as skin, limitations due to the method (stationarity, window size) prevent the use of Fourier analysis. Therefore, the authors have developed a new technique, based on the “centroid algorithm” using auto-regressive modeling. It allows a quasi-local mean-frequency estimation. Preliminary experimental results obtained at 20 MHz on porcine skin in vitro are shown. Acquisition in double transmission allows the measurement of the frequency dependence of attenuation. From the echographic signal, conventional B-scan images show two different regions that have been histologically identified to be the dermis and the hypodermis, respectively. Variations of the attenuation coefficient in the different skin layers (dermis, hypodermis) are detected and are reproducible on a set of nine skin samples. This study shows the feasibility of characterizing layered media using centroid and attenuation profiles