Learning from Projection to Reconstruction: A Deep Learning Reconstruction Framework for Sparse-View Phase Contrast Computed Tomography via Dual-Domain Enhancement

Abstract

Phase contrast computed tomography (PCCT) provides an effective non-destructive testing tool for weak absorption objects. Limited by the phase stepping principle and radiation dose requirement, sparse-view sampling is usually performed in PCCT, introducing severe artifacts in reconstruction. In this paper, we report a dual-domain (i.e., the projection sinogram domain and image domain) enhancement framework based on deep learning (DL) for PCCT with sparse-view projections. It consists of two convolutional neural networks (CNN) in dual domains and the phase contrast Radon inversion layer (PCRIL) to connect them. PCRIL can achieve PCCT reconstruction, and it allows the gradients to backpropagate from the image domain to the projection sinogram domain while training. Therefore, parameters of CNNs in dual domains are updated simultaneously. It could overcome the limitations that the enhancement in the image domain causes blurred images and the enhancement in the projection sinogram domain introduces unpredictable artifacts. Considering the grating-based PCCT as an example, the proposed framework is validated and demonstrated with experiments of the simulated datasets and experimental datasets. This work can generate high-quality PCCT images with given incomplete projections and has the potential to push the applications of PCCT techniques in the field of composite imaging and biomedical imaging.

Advances in Non-Destructive Testing Methods

“…The research topic of [27] is the generation of high-quality phase-contrast computed tomography images with given incomplete projections. The authors present the potential for expanding applications of PCCT techniques in the fields of composite and biomedical imaging and describe a two-domain (i.e., projection sinogram domain and image domain) deep-learning-based enhancement framework for PCCT with sparse projections.…”

Principle and Recent Development in Photonic Time-Stretch Imaging

“…The recent development of PTS imaging systems are mainly in four categories: first, PTS imaging systems combined with shorter wavelength bands for better imaging resolutions and rich applications [68][69][70][71][72][73]; second, PTS imaging systems with faster speed [21,22,[26][27][28][29]57,58,74,79]; third, PTS imaging systems combined with data compression [11][12][13][14][15][16][17][18][75][76][77][78]; and the last, PTS imaging systems combined with deep learning for target classification [23][24][25]32,66].…”

“…In order to improve the sensitivity, 2D devices, such as charge-coupled devices (CCD) and complementary metal oxide semiconductors (CMOS), 1D devices, such as photomultiplier tubes (PMT), avalanche photodetector (APD), and infrared-coated photodetector (PD) [60,61] are applied to improve the sensitivity of the imaging system. In order to obtain more specific information on the imaging systems, more techniques such as fluorescence imaging with biomarkers [62][63][64], phase-contrast imaging with interferometric structure for transparent sample imaging [65,66], and polarization-sensitive imaging are employed [67].…”

“…Among the improvement of current imaging systems, improving temporal resolution attracts a large amount of attention and obtained fruitful achievement. The most sparking technique for improving temporal resolution for imaging systems is PTS [66][67][68][69][70][71][72][73][74]-a technique that encodes the spatial profile of sample imaging information into the temporal profile data for ultrafast imaging using the dispersive medium in both spatial and temporal domains based on the dispersive properties of broadband light. It was proposed and demonstrated by Goda et al in 2009 for the first time [59].…”

Optical projection tomography reconstruction with few views using highly-generalizable deep learning at sinogram domain

“…There are three main approaches: 1. end-to-end image enhancement in spatial domain, such as CNOPT,DRONE [ 17 , 18 ]; 2. completion of sinogram in sinogram domain, followed by reconstruction using traditional algorithms, such as LDCT [ 19 , 20 ]; 3. Simultaneous optimization of both domains, such as DRONE [ 21 , 22 ]. These three approaches can effectively reduce the streaking artifacts, but it has not been discussed which approach is optimal considering generalization and performance.…”