Evaluation of high-resolution spectral estimation techniques in tissue thermal characterization under noisy broadband conditions

The spectral analysis techniques has been widely used to extract information about certain signals and the process in which those are generated. Ultrasonic (US) signal acquired from biological tissues are a particular case of that. Spectral analysis has been used to extract information from US signal related to biological tissue condition, mainly related with the internal scattering structure distribution and regularity. High-resolution spectral analysis techniques are applied to determine pathological characteristics that could indicate the presence of disease in the tissue. Some small changes in harmonics location could be drastically affected by the presence of noise in the analyzed signal. In this work, an evaluation of two parametric HR spectral analysis techniques is performed, first applied to signals acquired from a biological tissue phantom in a temperature range from 21°C to 45°C. A second evaluation was performed, considering simulated US signals, which were added with three different levels of SNR, and then they were analyzed with two parametric methods, which present variations in the error estimated for different noise level conditions. But the general behavior is a smaller error when noise decreases.