MRI Acoustic Noise cancellation using CNN

None

Abstract

Presented in this paper is a revolutionary deep learning-based architecture for reducing the noise generated during Magnetic Resonance Imaging (MRI) scans. The proposed architecture differs from the usual adaptive algorithms used in Active Noise Control (ANC). In the present work, we are exploring the use of Deep Convolutional Artificial Networks to recognize advanced sounds. By applying the DL-NN to a 513-time segment, a 180-degree phase shift sample of the noise is generated. After computational simulation analyses were performed, experimental results show that performance in noise average power can be reduced by approximately 10 to 15 dB.