Project 1: Physics-Informed Deep Learning CT Image Restoration

A multi-input CNN is proposed that incorporates image quality characteristics, such as blur and noise kernels, as auxiliary inputs to enhance image restoration.

Publications (1)

• Yijie Yuan, Grace J. Gang, J. Webster Stayman, "Deep learning CT image restoration using system blur and noise models," J. Med. Imag. 12(1) 014003 (3 February 2025). https://doi.org/10.1117/1.JMI.12.1.014003

Project 2: Physics-Informed Deep Learning CT Radiomics Standardization

Differentiable radiomics approximation algorithms are developed for Histogram, GLCM, and GLRLM-based features, enabling end-to-end deep learning networks to restore more accurate feature representations under variable imaging conditions.

Publications (2)

• Yijie Yuan, Huay Din, Grace Hyun Kim, Michael McNitt-Gray, J. Webster Stayman, Grace J. Gang, "Recovery of GLRLM Features in Degraded Images Using Deep Learning and Image Property Models."
• Yijie Yuan, J. Webster Stayman, Grace J. Gang, "End-to-end deep learning restoration of GLCM features from blurred and noisy images," Proc. SPIE 12927, Medical Imaging 2024: Computer-Aided Diagnosis, 129271C (3 April 2024). https://doi.org/10.1117/12.3006205