Journal of X-Ray Science and Technology最新文献

Background: Recent studies have explored layered correction strategies, employing a slice-by-slice approach to mitigate the prominent limited-view artifacts present in reconstructed images from high-pitch helical CT scans. However, challenges persist in determining the angles, quantity, and sequencing of slices.

Objective: This study aims to explore the optimal slicing method for high pitch helical scanning 3D reconstruction. We investigate the impact of slicing angle, quantity, order, and model on correction effectiveness, aiming to offer valuable insights for the clinical application of deep learning methods.

Methods: In this study, we constructed and developed a series of data-driven slice correction strategies for 3D high pitch helical CT images using slice theory, and conducted extensive experiments by adjusting the order, increasing the number, and replacing the model.

Results: The experimental results indicate that indiscriminately augmenting the number of correction directions does not significantly enhance the quality of 3D reconstruction. Instead, optimal reconstruction outcomes are attained by aligning the final corrected slice direction with the observation direction.

Conclusions: The data-driven slicing correction strategy can effectively solve the problem of artifacts in high pitch helical scanning. Increasing the number of slices does not significantly improve the quality of the reconstruction results, ensuring that the final correction angle is consistent with the observation angle to achieve the best reconstruction quality.

 Industrial digital radiography (DR) images are essential for industrial inspections, but they often suffer from strong scatter, cross-talk, electronic noise, and other factors that affect image quality. The presence of non-zero mean noise and neighborhood correlation loss in 1D array scanning poses significant challenges for denoising. To enhance the denoising process of industrial DR images and address the issues of low resolution and noise, we propose an improved KBNet (iKBNet) that incorporates lightweight modifications and introduces novel elements to the original KBNet. The iKBNet introduces the Convolutional Block Attention Module (CBAM) to reduce the network's parameter count. Additionally, it utilizes the Structural Similarity Index (SSIM) loss as part of a composite loss function to improve denoising performance. The proposed method demonstrates superior denoising results, with image restoration quality metrics that surpass those of commonly used methods such as BM3D, ResNet, DnCNN, and the original KBNet. In practical applications with low-resolution transmission images, the iKBNet has produced satisfactory outputs. The results indicate that the iKBNet not only minimizes computational cost and enhances processing speed but also achieves better denoising results. This suggests the potential of iKBNet for processing noisy digital radiographic images in industrial settings. The iKBNet shows promise in improving the quality of industrial DR images affected by noise, offering a viable solution for industrial image processing needs.

Background: Coherent diffraction imaging (CDI) is an important lens-free imaging method. As a variant of CDI, ptychography enables the imaging of objects with arbitrary lateral sizes. However, traditional phase retrieval methods are time-consuming for ptychographic imaging of large-size objects, e.g., integrated circuits (IC). Especially when ptychography is combined with computed tomography (CT) or computed laminography (CL), time consumption increases greatly.

Objective: In this work, we aim to propose a new deep learning-based approach to implement a quick and robust reconstruction of ptychography.

Methods: Inspired by the strong advantages of the residual dense network for computer vision tasks, we propose a dense residual two-branch network (RDenPtycho) based on the ptychography two-branch reconstruction architecture for the fast and robust reconstruction of ptychography. The network relies on the residual dense block to construct mappings from diffraction patterns to amplitudes and phases. In addition, we integrate the physical processes of ptychography into the training of the network to further improve the performance.

Results: The proposed RDenPtycho is evaluated using the publicly available ptychography dataset from the Advanced Photon Source. The results show that the proposed method can faithfully and robustly recover the detailed information of the objects. Ablation experiments demonstrate the effectiveness of the components in the proposed method for performance enhancement.

Significance: The proposed method enables fast, accurate, and robust reconstruction of ptychography, and is of potential significance for 3D ptychography. The proposed method and experiments can resolve similar problems in other fields.

Background and objective: Optimization based image reconstruction algorithm is an advanced algorithm in medical imaging. However, the corresponding solving algorithm is challenging because the model is usually large-scale and non-smooth. This work aims to devise a simple and convergent solver for optimization model.

Methods: The alternating direction method of multipliers (ADMM) algorithm is a simple and effective solver of the optimization model. However, there always exists a sub-problem that has not close-form solution. One may use gradient descent algorithm to solve this sub-problem, but the step-size selection via line search is time-consuming. Or, one may use fast Fourier transform (FFT) to get a close-form solution if the sparse transform matrix is of special structure. In this work, we propose a fully linearized ADMM (FL-ADMM) algorithm that avoids line search to determine step-size and applies to sparse transform of any structure.

Results: We derive the FL-ADMM algorithm instances for three total variation (TV) models in 2D computed tomography (CT). Further, we validate and evaluate one FL-ADMM algorithm and explore how two important factors impact convergence rate. These studies show that the FL-ADMM algorithm may accurately solve the optimization model.

Conclusion: The FL-ADMM algorithm is a simple, effective, convergent and universal solver of optimization model in image reconstruction. Compared to the standard ADMM algorithm, the new algorithm does not need time-consuming step-size line-search or special demand to sparse transform. It is a rapid prototyping tool for optimization based image reconstruction.

Background: Accurately detecting a variety of lung abnormalities from heterogenous chest X-ray (CXR) images and writing radiology reports is often difficult and time-consuming.

Objective: To access the utility of a novel artificial intelligence (AI) system (MOM-ClaSeg) in enhancing the accuracy and efficiency of radiologists in detecting heterogenous lung abnormalities through a multi-reader and multi-case (MRMC) observer performance study.

Methods: Over 36,000 CXR images were retrospectively collected from 12 hospitals over 4 months and used as the experiment group and the control group. In the control group, a double reading method is used in which two radiologists interpret CXR to generate a final report, while in the experiment group, one radiologist generates the final reports based on AI-generated reports.

Results: Compared with double reading, the diagnostic accuracy and sensitivity of single reading with AI increases significantly by 1.49% and 10.95%, respectively (P <  0.001), while the difference in specificity is small (0.22%) and without statistical significance (P = 0.255). Additionally, the average image reading and diagnostic time in the experimental group is reduced by 54.70% (P <  0.001).

Conclusion: This MRMC study demonstrates that MOM-ClaSeg can potentially serve as the first reader to generate the initial diagnostic reports, with a radiologist only reviewing and making minor modifications (if needed) to arrive at the final decision. It also shows that single reading with AI can achieve a higher diagnostic accuracy and efficiency than double reading.

Background: UNet has achieved great success in medical image segmentation. However, due to the inherent locality of convolution operations, UNet is deficient in capturing global features and long-range dependencies of polyps, resulting in less accurate polyp recognition for complex morphologies and backgrounds. Transformers, with their sequential operations, are better at perceiving global features but lack low-level details, leading to limited localization ability. If the advantages of both architectures can be effectively combined, the accuracy of polyp segmentation can be further improved.

Methods: In this paper, we propose an attention and convolution-augmented UNet-Transformer Network (ACU-TransNet) for polyp segmentation. This network is composed of the comprehensive attention UNet and the Transformer head, sequentially connected by the bridge layer. On the one hand, the comprehensive attention UNet enhances specific feature extraction through deformable convolution and channel attention in the first layer of the encoder and achieves more accurate shape extraction through spatial attention and channel attention in the decoder. On the other hand, the Transformer head supplements fine-grained information through convolutional attention and acquires hierarchical global characteristics from the feature maps.

Results: mcU-TransNet could comprehensively learn dataset features and enhance colonoscopy interpretability for polyp detection.

Conclusion: Experimental results on the CVC-ClinicDB and Kvasir-SEG datasets demonstrate that mcU-TransNet outperforms existing state-of-the-art methods, showcasing its robustness.

Background: The development of photon-counting CT systems has focused on semiconductor detectors like cadmium zinc telluride (CZT) and cadmium telluride (CdTe). However, these detectors face high costs and charge-sharing issues, distorting the energy spectrum. Indirect detection using Yttrium Orthosilicate (YSO) scintillators with silicon photomultiplier (SiPM) offers a cost-effective alternative with high detection efficiency, low dark count rate, and high sensor gain.

Objective: This work aims to demonstrate the feasibility of the YSO/SiPM detector (DexScanner L103) based on the Multi-Voltage Threshold (MVT) sampling method as a photon-counting CT detector by evaluating the synthesis error of virtual monochromatic images.

Methods: In this study, we developed a proof-of-concept benchtop photon-counting CT system, and employed a direct method for empirical virtual monochromatic image synthesis (EVMIS) by polynomial fitting under the principle of least square deviation without X-ray spectral information. The accuracy of the empirical energy calibration techniques was evaluated by comparing the reconstructed and actual attenuation coefficients of calibration and test materials using mean relative error (MRE) and mean square error (MSE).

Results: In dual-material imaging experiments, the overall average synthesis error for three monoenergetic images of distinct materials is 2.53% ±2.43%. Similarly, in K-edge imaging experiments encompassing four materials, the overall average synthesis error for three monoenergetic images is 4.04% ±2.63%. In rat biological soft-tissue imaging experiments, we further predicted the densities of various rat tissues as follows: bone density is 1.41±0.07 g/cm3, adipose tissue density is 0.91±0.06 g/cm3, heart tissue density is 1.09±0.04 g/cm3, and lung tissue density is 0.32±0.07 g/cm3. Those results showed that the reconstructed virtual monochromatic images had good conformance for each material.

Conclusion: This study indicates the SiPM-based photon-counting detector could be used for monochromatic image synthesis and is a promising method for developing spectral computed tomography systems.

Background: Dynamic cerebral perfusion CT (DCPCT) can provide valuable insight into cerebral hemodynamics by visualizing changes in blood within the brain. However, the associated high radiation dose of the standard DCPCT scanning protocol has been a great concern for the patient and radiation physics. Minimizing the x-ray exposure to patients has been a major effort in the DCPCT examination. A simple and cost-effective approach to achieve low-dose DCPCT imaging is to lower the x-ray tube current in data acquisition. However, the image quality of low-dose DCPCT will be degraded because of the excessive quantum noise.

Objective: To obtain high-quality DCPCT images, we present a statistical iterative reconstruction (SIR) algorithm based on penalized weighted least squares (PWLS) using adaptive prior image constrained total generalized variation (APICTGV) regularization (PWLS-APICTGV).

Methods: APICTGV regularization uses the precontrast scanned high-quality CT image as an adaptive structural prior for low-dose PWLS reconstruction. Thus, the image quality of low-dose DCPCT is improved while essential features of targe image are well preserved. An alternating optimization algorithm is developed to solve the cost function of the PWLS-APICTGV reconstruction.

Results: PWLS-APICTGV algorithm was evaluated using a digital brain perfusion phantom and patient data. Compared to other competing algorithms, the PWLS-APICTGV algorithm shows better noise reduction and structural details preservation. Furthermore, the PWLS-APICTGV algorithm can generate more accurate cerebral blood flow (CBF) map than that of other reconstruction methods.

Conclusions: PWLS-APICTGV algorithm can significantly suppress noise while preserving the important features of the reconstructed DCPCT image, thus achieving a great improvement in low-dose DCPCT imaging.

Background: Deformable image registration (DIR) plays an important part in many clinical tasks, and deep learning has made significant progress in DIR over the past few years.

Objective: To propose a fast multiscale unsupervised deformable image registration (referred to as FMIRNet) method for monomodal image registration.

Methods: We designed a multiscale fusion module to estimate the large displacement field by combining and refining the deformation fields of three scales. The spatial attention mechanism was employed in our fusion module to weight the displacement field pixel by pixel. Except mean square error (MSE), we additionally added structural similarity (ssim) measure during the training phase to enhance the structural consistency between the deformed images and the fixed images.

Results: Our registration method was evaluated on EchoNet, CHAOS and SLIVER, and had indeed performance improvement in terms of SSIM, NCC and NMI scores. Furthermore, we integrated the FMIRNet into the segmentation network (FCN, UNet) to boost the segmentation task on a dataset with few manual annotations in our joint leaning frameworks. The experimental results indicated that the joint segmentation methods had performance improvement in terms of Dice, HD and ASSD scores.

Conclusions: Our proposed FMIRNet is effective for large deformation estimation, and its registration capability is generalizable and robust in joint registration and segmentation frameworks to generate reliable labels for training segmentation tasks.