Ultrasound 3D imaging system utilizing bat-inspired artificial neural networks

Abstract

In this paper, a bat-inspired high-resolution ultrasound 3D imaging system is presented. Live bats demonstrate that the properly used ultrasound can be used to perceive 3D space. With this in mind, a neural network referred to as a Bat-G network is implemented to reconstruct the 3D representation of target objects from the hyperbolic FM (HFM) chirped ultrasonic echoes. The Bat-G network consists of an encoder emulating a bat's central auditory pathway, and a 3D graphical visualization decoder. For the acquisition of the ultrasound data, a custom-made Bat-I sensor module is used. The Bat-G network shows the uniform 3D reconstruction results and achieves precision, recall, and F1-score of 0.896, 0.899, and 0.895, respectively. The experimental results demonstrate the implementation feasibility of a high-resolution non-optical sound-based imaging system being used by live bats. And a noise-immune Bat-inspired Graphical visualization network Guided by the radiated ultrasonic call (Bat-G2 net) that can reconstruct 3D shapes of a target from ultrasonic echoes under low SNR conditions like live bats is presented. Bat-G2 net is implemented by emulating robust noise-resilient bat's auditory system that process echoes along with the highly correlated radiated ultrasonic call (RUC). In order to extract the information contained in the RUC robustly and effectively, two implementation ideas have been applied to the Bat-G2 net: (1) RUC-guided attention (2) non-local attention. The Bat-G2 net is trained with ECHO-4CH dataset acquired by a custom-made Bat-I sensor. Noise-resistant property of the Bat-G2 net is demonstrated by the outstanding reconstruction result for the noise added sensory input as compared to the result of current state-of-the-art ultrasonic image reconstruction network. This study clearly demonstrates the implementation feasibility of the new modality of “seeing by hearing” in more practical environments.