Simulating Patho-realistic Ultrasound Images using Deep Generative Networks with Adversarial Learning
share
Ultrasound imaging makes use of backscattering of waves during their interaction with scatterers present in biological tissues. Simulation of synthetic ultrasound images is a challenging problem on account of inability to completely model various factors of which some include intra-/inter scanline interference, transducer to surface coupling, artifacts on transducer elements, inhomogeneous shadowing and nonlinear attenuation. While various approaches to ultrasound simulation has been developed, approaches that produce patho-realistic images typically solve wave space equations making it computationally expensive and slow to operate. We propose a generative adversarial network (GAN) inspired approach for fast simulation of patho-realistic ultrasound images. We apply the framework to intravascular ultrasound (IVUS) simulation. A stage 0 simulation is done from the echogenicity map of the tissue obtained from the ground truth label of ultrasound image using an off the shelf pseudo B-mode ultrasound image simulator . The images obtained are adversarially refined using stacked GAN. The stage I GAN generates low resolution images from the images generated by the initial simulation. The stage II GAN further refines the output of the stage I GAN and generates high resolution images which are patho-realistic. We demonstrate that the network generates realistic appearing images evaluated with a visual Turing test indicating an equivocal confusion in discriminating simulated from real. We also quantify the shift in tissue specific intensity distributions of the real and simulated images to prove their similarity.
READ FULL TEXT