Model-Based Iterative Reconstruction (MBIR) has shown promising results in clinical studies as it allows significant dose reduction during CT scans while maintaining the diagnostic image quality. MBIR improves the image quality over analytical reconstruction by modeling both the sensor (e.g., forward model) and the image being reconstructed (e.g., prior model). While the forward model is typically based on the physics of the sensor, accurate prior modeling remains a challenging problem. In order to tackle this challenge, we use the Plug-and-Play (PnP) framework, which decouples the forward and prior models in the optimization framework and allows any denoising algorithm to serve as a prior model. We then adopt the state-of-the-art deep residual learning approach for training the image DNN denoiser using medical CT images. Experimental results on real CT scans demonstrate that our PnP MBIR with deep residual learning prior significantly reduces the noise and artifacts compared to analytical reconstruction and standard MBIR with an MRF prior.

This presentation is part of Minisymposium “MS24 - Data-driven approaches in imaging science (3 parts)”
organized by: Jae Kyu Choi (Institute of Natural Sciences, Shanghai Jiao Tong University) , Chenglong Bao (Yau Mathematical Sciences Center, Tsinghua University) .

Authors:

Charles Bouman (Purdue University)

Keywords: