An Improved Iterative Neural Network for High-Quality Image-Domain Material Decomposition in Dual-Energy CT

by   Zhipeng Li, et al.

Dual-energy computed tomography (DECT) has been widely used in many applications that need material decomposition. Image-domain methods directly decompose material images from high- and low-energy attenuation images, and thus, are susceptible to noise and artifacts on attenuation images. To obtain high-quality material images, various data-driven methods have been proposed. Iterative neural network (INN) methods combine regression NNs and model-based image reconstruction algorithm. INNs reduced the generalization error of (noniterative) deep regression NNs, and achieved high-quality reconstruction in diverse medical imaging applications. BCD-Net is a recent INN architecture that incorporates imaging refining NNs into the block coordinate descent (BCD) model-based image reconstruction algorithm. We propose a new INN architecture, distinct cross-material BCD-Net, for DECT material decomposition. The proposed INN architecture uses distinct cross-material convolutional neural network (CNN) in image refining modules, and uses image decomposition physics in image reconstruction modules. The distinct cross-material CNN refiners incorporate distinct encoding-decoding filters and cross-material model that captures correlations between different materials. We interpret the distinct cross-material CNN refiner with patch perspective. Numerical experiments with extended cardiactorso (XCAT) phantom and clinical data show that proposed distinct cross-material BCD-Net significantly improves the image quality over several image-domain material decomposition methods, including a conventional model-based image decomposition (MBID) method using an edge-preserving regularizer, a state-of-the-art MBID method using pre-learned material-wise sparsifying transforms, and a noniterative deep CNN denoiser.


page 1

page 6

page 7

page 9

page 15

page 16

page 17

page 18


BCD-Net for Low-dose CT Reconstruction: Acceleration, Convergence, and Generalization

Obtaining accurate and reliable images from low-dose computed tomography...

DECT-MULTRA: Dual-Energy CT Image Decomposition With Learned Mixed Material Models and Efficient Clustering

Dual energy computed tomography (DECT) imaging plays an important role i...

Deep Learning for Material Decomposition in Photon-Counting CT

Photon-counting CT (PCCT) offers improved diagnostic performance through...

Deep BCD-Net Using Identical Encoding-Decoding CNN Structures for Iterative Image Recovery

In "extreme" computational imaging that collects extremely undersampled ...

A Splitting-Based Iterative Algorithm for GPU-Accelerated Statistical Dual-Energy X-Ray CT Reconstruction

When dealing with material classification in baggage at airports, Dual-E...

Multiresolution Neural Networks for Imaging

We present MR-Net, a general architecture for multiresolution neural net...

Momentum-Net: Fast and convergent iterative neural network for inverse problems

Iterative neural networks (INN) are rapidly gaining attention for solvin...

Please sign up or login with your details

Forgot password? Click here to reset