Improving SNR of Low-Frequency Ultrasound Thoracic Signal Processing Based on Music Algorithm and MEMD Wavelet Thresholding

A common tool in biomedical imaging is ultrasound. For good resolution, high-frequency ultrasound operating at MHz is typically employed. However, because air inclusions in the thorax greatly scatter and reflect ultrasound, it cannot penetrate the human thorax at this band. This restriction can be overcome by low-frequency ultrasonography, which can also obtain important data from deep within the thorax. In order to improve the signal quality of low-frequency ultrasound imaging, this work offers a new signal processing technique that lowers the mean square error (MSE) and increases the signal-to-noise ratio (SNR). In particular, we provide a method that combines the Multiple Signal Classification (MUSIC) algorithm with the wavelet thresholding approach based on Multi-Variate Empirical Mode Decomposition (MEMD). For additional noise reduction, the MEMD-based wavelet threshold denoising technique is applied after the low-frequency ultrasound signal has been pre-processed using the MUSIC algorithm. We evaluate our method's efficacy using simulation and experimental studies. The simulated results show significant gains in low-frequency ultrasound signal quality, with an SNR enhancement and an MSE drop. Additionally, an SNR improvement range and a matching MSE decrease of are obtained from the trial. These results, which are consistent with the simulation results, validate the effectiveness of our suggested approach as a cost-effective means of enhancing the quality of low-frequency ultrasound transmissions.

Keywords: Ultrasound, Multiple Signal Classification (MUSIC) Algorithm, Multi-variate empirical mode decomposition (MEMD), Signal-to-Noise Ratio (SNR), Mean Square Error (MSE).