<p>Ultrasound imaging is one of the most widely used non-destructive testing<br />methods. The transducer emits pulses that travel through the imaged samples<br />and are reflected by echo-forming impedance. The resulting ultrasonic signals<br />usually contain noise. Most of the traditional noise reduction algorithms<br />require high skills and prior knowledge of noise distribution, which has a<br />crucial impact on their performances. As a result, these methods generally<br />yield a loss of information, significantly influencing the final data and deeply<br />limiting both sensitivity and resolution of imaging devices in medical and<br />industrial applications. In the present study, a denoising method based on an<br />attention-gated convolutional autoencoder is proposed to fill this gap. To<br />evaluate its performance, the suggested protocol is compared to widely used<br />methods such as butterworth filtering (BF), discrete wavelet transforms<br />(DWT), principal component analysis (PCA), and convolutional autoencoder<br />(CAE) methods. Results proved that better denoising can be achieved<br />especially when the original signal-to-noise ratio (SNR) is very low and the<br />sound waves’ traces are distorted by noise. Moreover, the initial SNR was<br />improved by up to 30 dB and the resulting Pearson correlation coefficient was<br />maintained over 99% even for ultrasonic signals with poor initial SNR.</p>