Image Denoising and Ring Artifacts Removal for Spectral CT via Deep Neural Network

Xiaojie Lv,Xuezhi Ren,Chengyu Fan,Peng Feng,Biao Wei,Mi Zhou,Xiaodong Guo,Zourong Long,Peng He

doi:10.1109/access.2020.3044708

Abstract

The spectral computed tomography (CT) based on photon counting detectors can collect the incident photons with different energy ranges. However, due to the low photon counts in narrow energy bin and the unhomogeneous response problem of detector cells, there are severe noise and ring artifacts in reconstructed spectral CT images. We proposed an image denoising and ring artifacts removal method via improved Fully Convolutional Pyramid Residual Network (FCPRN). In our study, we scanned a mouse specimen with spectral CT based on photon counting detector, and reconstructed mouse CT images as data set. Then we use the data set to train our network for image denoising and ring artifacts removal. Experimental results demonstrated that the proposed method could reduce noise and suppress ring artifacts of spectral CT images concurrently in different energy ranges. And the performance of the FCPRN is better than that of some networks for CT image denoising.

Highlights

Spectral computed tomography (CT) based on photon counting detector can identify absorption features in the available range of photon energies, which has a stronger capability to distinguish different materials [1]
To suppress noise and remove ring artifacts for spectral CT images simultaneously, we proposed an image denoising and ring artifacts removal method via fully convolutional pyramid residual network (FCPRN)
The skip connection could transmit the feature information to the deep layers, which is helpful for recovering the resolution of the feature map, recognizing every pixel in the image and extracting noise and artifacts features

Summary

Introduction

Spectral CT based on photon counting detector can identify absorption features in the available range of photon energies, which has a stronger capability to distinguish different materials [1]. The unhomogeneous response of detector cells can generate ring artifacts [4] in reconstructed images. Noise and artifacts in reconstructed spectral CT images will seriously affect diagnosis and recognition. Some research methods have been presented to suppress noise in CT images. Zhang et al [5] proposed a CT image denoising scheme based on dictionary learning method. Wu et al [6] studied a parallelized non-local means denoising method to suppress noise of CT images. Kang et al [12] proposed a deep CNN framework to solve CT image denoising problems, which combines a deep convolution neural network with a directional wavelet approach. Shan et al [16, 17] proposed a modularized neural network for LDCT (low-dose CT), which could effectively suppress image noise

Methods

Results

Conclusion