Journal of X-Ray Science and Technology最新文献

Cone beam computed tomography (CBCT) is increasingly used in clinical settings, with the radiation dose incurred during X-ray acquisition emerging as a critical concern. Traditional algorithms for reconstructing high-quality CBCT images typically necessitate hundreds of X-ray projections, prompting a shift towards sparse-view CBCT reconstruction as a means to minimize radiation exposure. A novel approach, leveraging the Neural Attenuation Field (NAF) based on neural radiation field algorithms, has recently gained traction. This method offers rapid and promising CBCT reconstruction outcomes using a mere 50 views. Nonetheless, NAF tends to overlook the inherent structural properties of projected images, which can lead to shortcomings in accurately capturing the structural essence of the object being imaged. To address these limitations, we introduce an enhanced method: Regularization Neural Attenuation Fields (RNAF). Our approach includes two key innovations. First, we implement a hash coding regularization technique designed to retain low-frequency details within the reconstructed images, thereby preserving essential structural information. Second, we incorporate a Local Patch Global (LPG) sampling strategy. This method focuses on extracting local geometric details from the projection image, ensuring that the intensity variations in randomly sampled X-rays closely mimic those in the actual projection image. Comparative analyses across various body parts (Chest, Jaw, Foot, Abdomen, Knee) reveal that RNAF substantially outperforms existing algorithms. Specifically, its reconstruction quality exceeds that of previous NeRF-based, optimization-based, and analysis algorithms by margins of at least 2.09 dB, 3.09 dB, and 13.84 dB respectively. This significant enhancement in performance underscores the potential of RNAF as a groundbreaking solution in the realm of CBCT imaging, offering a path towards achieving high-quality reconstructions with reduced radiation exposure. Our implementation is publically available at https://github.com/springXIACJ/FRNAF.

This research introduces a Multistage-Vision Transformer (Multistage-ViT) model for precisely classifying various lung diseases using chest radiographic (CXR) images. The dataset in the proposed method includes four classes: Normal, COVID-19, Viral Pneumonia and Lung Opacity. This model demonstrates its efficacy on imbalanced and balanced datasets by enhancing classifier accuracy through deep feature extraction. It integrates backbone models with the ViT architecture, creating rigorously hybrid configurations compared to their standalone counterparts. These hybrid models utilize optimized features for classification, significantly improving their performance. Notably, the multistage-ViT model achieved accuracies of 99.93% on an imbalanced dataset and 99.97% on a balanced dataset using the InceptionV3 combined with the ViT model. These findings highlight the superior accuracy and robustness of multistage-ViT models, underscoring their potential to enhance lung disease classification through advanced feature extraction and model integration techniques. The proposed model effectively demonstrates the benefits of employing ViT for deep feature extraction from CXR images.

The objective of this study is to analyse and validate the distribution and missing regions in the frequency domain space of the projection information obtained from Cone-beam Computed Laminography (CBCL) scanned samples. Furthermore, the aim is to establish a frequency domain information distribution and missing model for CBCL. This paper employs the Fourier slice theorem to ascertain the spatial region of the frequency domain wherein the CBCL projection information is situated. To this end, the geometrical structure of the CL system and the spatial propagation characteristics of the cone-beam rays are subjected to analysis. Furthermore, the veracity of the model for the missing information in the CBCL frequency domain is validated through an iterative reconstruction process, whereby different regions of the frequency domain space are reconstructed through an iterative reconstruction algorithm that takes only the projection information as a constraint. The CBCL frequency domain missing information model can be employed as a priori information in the frequency domain space to facilitate further optimisation and improvement of image reconstruction.

Computed tomography (CT) image reconstruction is faced with challenge of balancing image quality and radiation dose. Recent unrolled optimization methods address low-dose CT image quality issues using convolutional neural networks or self-attention mechanisms as regularization operators. However, these approaches have limitations in adaptability, computational efficiency, or preservation of beneficial inductive biases. They also depend on initial reconstructions, potentially leading to information loss and error propagation. To overcome these limitations, Kernel Basis Attention Primal-Dual Network (KBA-PDNet) is proposed. The method unrolls multiple iterations of the proximal primal-dual optimization process, replacing traditional proximal operators with Kernel Basis Attention (KBA) modules. This design enables direct training from raw measurement data without relying on preliminary reconstructions. The KBA module achieves adaptability by learning and dynamically fusing kernel bases, generating customized convolution kernels for each spatial location. This approach maintains computational efficiency while preserving beneficial inductive biases of convolutions. By training end-to-end from raw projection data, KBA-PDNet fully utilizes all original information, potentially capturing details lost in preliminary reconstructions. Experiments on simulated and clinical datasets demonstrate that KBA-PDNet outperforms existing approaches in both image quality and computational efficiency.

In recent years, spectral computed tomography (CT) has attracted extensive attention. The purpose of this study is to achieve a low-cost and fast energy spectral CT reconstruction algorithm by implementing multi-limited-angle scans. General spectral CT projection data are collected over a full-angular range of 360 degrees. We simulate multi-source spectral CT by using a pair of X-ray source/detector. To speed up scanning, multi-limited-angle scanning was used in each energy channel. On this basis, an average image induced relative total variation (Aii-RTV) with multi-limited-angle spectral CT image reconstruction model is proposed. The iterative algorithm is used to solve Aii-RTV. Before iteration, the weighted average projection data of the multi-limited-angle energy spectral is carried out. In each step of the iterative algorithm flow is as follows: First, the relative total variation (RTV) reconstruction model is used to reconstruct the average image using average projection data. Then, the partial derivative of the average image is used to calculate the inherent variation in RTV model due to the integrity of the average image, and take its reciprocal as the weight coefficient of the windowing total variation of each energy channel reconstruction image. Finally, the average energy image is used to guide the multi-limited-angle projection data to reconstruct the image of each energy channel so as to suppress the limited-angle artifact of each energy channel image. In addition, we also discuss the influence of parameter selection on reconstructed image quality, which is important for regularization model. Through the reconstruction of multi-limited-angle spectral CT projection data, quantitative results and reconstructed images show that our algorithm has better performance than prior image constrained compressed sensing (PICCS) and RTV. The average PSNR of our reconstruction results in different channels was 35.6273, 4.533 and 2.301 higher than RTV (31.0943) and PICCS (33.3263), respectively.

Knee arthritis is a prevalent joint condition that affects many people worldwide. Early detection and appropriate treatment are essential to slow the disease's progression and enhance patients' quality of life. In this study, various machine learning and deep learning algorithms were used to detect knee arthritis. The machine learning models included k-NN, SVM, and GBM, while DenseNet, EfficientNet, and InceptionV3 were used as deep learning models. Additionally, YOLOv8 classification models (YOLOv8n-cls, YOLOv8s-cls, YOLOv8m-cls, YOLOv8l-cls, and YOLOv8x-cls) were employed. The "Annotated Dataset for Knee Arthritis Detection" with five classes (Normal, Doubtful, Mild, Moderate, Severe) and 1650 images were divided into 80% training, 10% validation, and 10% testing using the Hold-Out method. YOLOv8 models outperformed both machine learning and deep learning algorithms. k-NN, SVM, and GBM achieved success rates of 63.61%, 64.14%, and 67.36%, respectively. Among deep learning models, DenseNet, EfficientNet, and InceptionV3 achieved 62.35%, 70.59%, and 79.41%. The highest success was seen in the YOLOv8x-cls model at 86.96%, followed by YOLOv8l-cls at 86.79%, YOLOv8m-cls at 83.65%, YOLOv8s-cls at 80.37%, and YOLOv8n-cls at 77.91%.

X-ray computed tomography (XCT) is increasingly being used for the measurement and inspection of large dense metallic engineering components. When scanning such components, the quality of the data is degraded by the presence of scattered radiation. In this work, the performance of a focused 2D anti-scatter grid (ASG) is investigated for scanning samples made from cobalt chrome and Inconel on a 450 kV cone-beam XCT system. The devised scatter correction method requires one additional scan of the sample, and for projections to be algorithmically processed prior to reconstruction. The results show that the ASG based scatter correction method increases the contrast-to-noise of the data by 14.5% and 61.5% for the cobalt chrome and Inconel samples, respectively. Furthermore, the method increases edge sharpness by 6% and 16.9% for outer and inner edges, respectively.

Background: Chest X-rays are an essential diagnostic tool for identifying chest disorders because of its high sensitivity in detecting pathological anomalies in the lungs. Classification models based on conventional Convolutional Neural Networks (CNNs) are adversely affected due to their localization bias. Objective: In this paper, a new Multi-Axis Transformer based U-Net with Class Balanced Ensemble (MaxTU-CBE) is proposed to improve multi-label classification performance. Methods: This may be the first attempt to simultaneously integrate the benefits of hierarchical Multi-Axis Transformer into the encoder and decoder of the traditional U-shaped structure for improving the semantic segmentation superiority of lung image. Results: A key element of MaxTU-CBE is the Contextual Fusion Engine (CFE), which uses the self-attention mechanism to efficiently create global interdependence between features of various scales. Also, deep CNN incorporate ensemble learning to address the issue of class unbalanced learning. Conclusions: According to experimental findings, our suggested MaxTU-CBE outperforms the competing BiDLSTM classifier by 1.42% and CBIR-CSNN techniques by 5.2% in multi-label classification performance.

BackgroundDiffusion-weighted imaging (DWI) is an important technique to study brain microstructure. However, diffusion-weighted (DW) images suffer from severe low signal-to-noise ratio (SNR) problem, affecting subsequent diffusion analysis.ObjectiveThe goal of this paper is to develop advanced DWI denoising technique to effectively reduce noise while improving the accuracy and reliability of subsequent diffusion model fitting and diffusion analysis, thereby facilitating the research and analysis of brain science.MethodsWe propose a new method for denoising DW images based on patch-matching with higher-order singular value decomposition (HOSVD) by combined with the variance-stabilizing transformation technique. It starts with introducing a novel non-local mean algorithm as a prefiltering stage, and then denoises the noisy data using a local HOSVD algorithm based on the HOSVD bases learned from prefiltered images.ResultsExperiments are performed on simulation, HCP and in vivo brain DWI datasets. Results show that the proposed method significantly reduces spatially invariant and variant noise, improving the most reliable diffusion analysis compared with the different denoising methods.ConclusionsThe proposed method achieves state-of-the-art performance which can improve image quality and enable accurate diffusion analysis.

BackgroundLymph node metastasis (LNM) is a poor prognostic predictor and is highly correlated with local recurrence in rectal cancer patients.ObjectiveTo investigate the value of radiomics from dual-energy CT-derived iodine maps for the preoperative prediction of LNM in rectal cancer patients.MethodsA total of 176 patients were enrolled in this study (training group, n = 123; validation group, n = 53). A radiomic signature was constructed via support vector machine (SVM) modeling. Seven models, including a clinical feature model (Model 1), an arterial model (Model 2), a venous model (Model 3), an arterial-venous model (Model 4), an arterial-clinical model (Model 5), a venous-clinical model (Model 6) and an arterial-venous-clinical model (Model 7), were established via logistic regression modeling. Diagnostic performance was assessed via receiver operating characteristic (ROC) curves.ResultsTumor location and carcinoembryonic antigen levels were used to construct Model 1 (training group, AUC [area under the ROC curve] = 0.721, 95% CI [confidence intervals], 0.630-0.813; validation group, AUC = 0.729, 95% CI, 0.593-0.865). Model 6 and Model 7 further improved the discriminatory performance in the training (AUC = 0.850 and 0.869, 95% CI, 0.782-0.919 and 0.807-0.932, respectively; p = 0.250) and validation groups (AUC = 0.780 and 0.716, 95% CI, 0.653-0.906 and 0.576-0.856, respectively; p = 0.115). Moreover, decision curve analysis revealed a greater net benefit with Model 6.ConclusionsThe combination of radiomic features based on dual-energy CT-derived iodine maps and clinical features provides better diagnostic performance for predicting LNM in rectal cancer patients.