A novel fusion method for X-ray phase contrast imaging based on fast adaptive bidimensional empirical mode decomposition.

BACKGROUNDS: X-ray phase contrast imaging (XPCI) can separate the attenuation, refraction, and scattering signals of the object. The application of image fusion enables the concentration of distinctive information into a single image. Some methods have been applied in XPCI field, but wavelet-based decomposition approaches often result in loss of original data. OBJECTIVE: To explore the application value of a novel image fusion method for XPCI system and computed tomography (CT) system. METHODS: The means of fast adaptive bidimensional empirical mode decomposition (FABEMD) is considered for image decomposition to avoid unnecessary information loss. A parameter δ is proposed to guide the fusion of bidimensional intrinsic mode functions which contain high-frequency information, using a pulse coupled neural network with morphological gradients (MGPCNN). The residual images are fused by the energy attribute fusion strategy. Image preprocessing and enhancement are performed on the result to ensure its quality. The effectiveness of other image fusion methods has been compared, such as discrete wavelet transforms and anisotropic diffusion fusion. RESULTS: The δ-guided FABEMD-MGPCNN method achieved either the first or second position in objective evaluation metrics with biological samples, as compared to other image fusion methods. Moreover, comparisons are made with other fusion methods used for XPCI. Finally, the proposed method applied in CT show expected results to retain the feature information. CONCLUSIONS: The proposed δ-guided FABEMD-MGPCNN method shows potential feasibility and superiority over traditional and recent image fusion methods for X-ray differential phase contrast imaging and computed tomography systems.

Moiré artifacts reduction in Talbot-Lau X-ray phase contrast imaging using a three-step iterative approach.

Talbot-Lau X-ray phase contrast imaging is a promising technique in biological imaging since it can provide absorption, differential phase contrast, and dark-field images simultaneously. However, high accuracy motorized translation stages and high stability of the imaging system are needed to avoid moiré artifacts in the reconstructed images. In this work, the effects of the stepping errors and the dose fluctuations on the transmission, differential phase contrast, and dark-field images are theoretically derived and systematically summarized. A novel three-step iterative method is designed for image reconstruction in Talbot-Lau interferometry with phase-stepping errors and dose fluctuations. Phase distributions, phase-stepping errors, and dose fluctuation coefficients are iteratively updated via the least square method until the convergence criteria are met. Moiré artifacts are mostly reduced via the proposed method in both the numerical simulations and experiments. The reconstructed images are highly coincident with the ground truth, which verifies the high accuracy of this method. The proposed algorithm is also compared with other moiré artifacts reduction algorithms, which further demonstrates the high precision of this algorithm. This work is beneficial for reducing the strict requirements for the hardware system in the conventional Talbot-Lau interferometry, such as the high accuracy motorized stages and the X-ray tube with high stability, which is significant for advancing the X-ray phase contrast imaging towards the practical medical applications.

Phase retrieval for X-ray differential phase contrast radiography with knowledge transfer learning from virtual differential absorption model.

Grating-based X-ray phase contrast radiography and computed tomography (CT) are promising modalities for future medical applications. However, the ill-posed phase retrieval problem in X-ray phase contrast imaging has hindered its use for quantitative analysis in biomedical imaging. Deep learning has been proved as an effective tool for image retrieval. However, in practical grating-based X-ray phase contrast imaging system, acquiring the ground truth of phase to form image pairs is challenging, which poses a great obstacle for using deep leaning methods. Transfer learning is widely used to address the problem with knowledge inheritance from similar tasks. In the present research, we propose a virtual differential absorption model and generate a training dataset with differential absorption images and absorption images. The knowledge learned from the training is transferred to phase retrieval with transfer learning techniques. Numerical simulations and experiments both demonstrate its feasibility. Image quality of retrieved phase radiograph and phase CT slices is improved when compared with representative phase retrieval methods. We conclude that this method is helpful in both X-ray 2D and 3D imaging and may find its applications in X-ray phase contrast radiography and X-ray phase CT.

A multimodal image feature extraction method for x-ray grating phase contrast computed tomography based on monogenic signal.

Traditional computed tomography (CT) based on x-ray absorption imaging has made great progress in clinical medicine, and CT combined with x-ray phase contrast imaging (XPCI) technology has become a new research hotspot in recent years. XPCI can separate the attenuation, refraction, and scattering signals of the object and retrieve three types of feature images known as absorption contrast image, differential phase contrast image, and dark field contrast image. However, the quality of CT images is always degraded due to noise and reconstruction artifacts, which makes feature recognition methods for CT images necessary. Most of the existing CT image recognition algorithms are focused on AC-CT images, with little attention paid to other contrast images. Herein, a new method is proposed, named the variable kernel multi-scale adaptive monogenic signal phase consistency model (VK-MA PC model), which constructs monogenic signals with corresponding filters according to the characteristics of different contrast images. The model obtains better image features by using multi-scale analysis and optional pre-decomposition, which make images decomposed into different levels. Experiments on 4D extended cardiac-torso (XCAT) human body simulation data and laboratory fish XPCI-CT data demonstrate the potential applicability of the VK-MA PC model in the field of XPCI-CT.

Tri-directional x-ray phase contrast multimodal imaging using one hexagonal mesh modulator.

Objective. X-ray phase contrast imaging is a promising technique for future clinical diagnostic as it can provide enhanced contrast in soft tissues compared to traditional x-ray attenuation-contrast imaging. However, the strict requirements on the x-ray coherence and the precise alignment of optical elements limit its applications towards clinical use. To solve this problem, mesh-based x-ray phase contrast imaging method with one hexagonal mesh is proposed for easy alignment and better image visualization.Approach. The mesh produces structured illuminations and the detector captures its distortions to reconstruct the absorption, differential phase contrast (DPC) and dark-field (DF) images of the sample. In this work, we fabricated a hexagonal mesh to simultaneously retrieve DPC and DF signals in three different directions with single shot. A phase retrieval algorithm to obtain artifacts-free phase from DPC images with three different directions is put forward and false color dark-field image is also reconstructed with tri-directional images. Mesh-shifting method based on this hexagonal mesh modulator is also proposed to reconstruct images with better image quality at the expense of increased dose.Main results. In numerical simulations, the proposed hexagonal mesh outperforms the traditional square mesh in image evaluation metrics performance and false color visualization with the same radiation dose. The experimental results demonstrate its feasiblity in real imaging systems and its advantages in quantitive imaging and better visualization. The proposed hexagonal mesh is easy to fabricate and can be successfully applied to x-ray source with it spot size up to 300µm.Significance. This work opens new possibilities for quantitative x-ray non-destructive imaging and may also be instructive for research fields such as x-ray structured illumination microscopy (SIM), x-ray spectral imaging and x-ray phase contrast and dark-field computed tomography (CT).