CT Reconstruction from Ultrasound Images of Breast Cancer using GANs

Research Assistant (Collaboration with TATA Cancer Research Hospital) at the Medical Deep Learning and Artificial Intelligence Lab, IIT Bombay, Guide: Prof. Amit Sethi

Summary: Ultrasound (US) imaging is cost-effective, safe and real-time, but it contains noise and artifacts. This project aimed at providing surgeons with mammogram-like image quality in real-time from by modelling ultrasound noise.

Paper submitted to the Ultrasonics Journal. 1 review round completed.

Paper link

  • Pre-processed CTs to speed of sound images to simulate US through wave interference equations using the Stride module
  • Applied Fourier Domain Adaptation (FDA) to enhance the quality of simulated images using authentic ultrasound images
  • Utilized CycleGANs and Pix2pix to generate CT scans from simulated ultrasound images, achieving an MSE of 0.008