Available at: https://digitalcommons.calpoly.edu/theses/2962
Date of Award
12-2024
Degree Name
MS in Electrical Engineering
Department/Program
Electrical Engineering
College
College of Engineering
Advisor
Wayne Pilkington
Advisor Department
Electrical Engineering
Advisor College
College of Engineering
Abstract
This paper focuses on improving speckle reduction and denoising performance in ultrasound images by leveraging neural networks. A dataset of simulated ultrasound images was created using Field-II simulation software based on CT scan images, to create clean and noisy image pairs. Various Convolutional Neural Network models based on U-Net and generative adversarial networks were developed and tested. Peak Signal-to-Noise-Ratio (PSNR) and Structural Similarity Index Measurement (SSIM) were used as metrics along with qualitative assessment. Results show that our tuned U-Net generator network outperformed traditional filtering such as Lee and BM3D.