Journal of X-Ray Science and Technology最新文献

BackgroundThe supporting structure of high-temperature gas-cooled reactors (HTGR) comprises over 3000 carbon/graphite components, necessitating computed tomography (CT) non-destructive testing before operational deployment as per reactor technical specifications. However, CT images are frequently marred by severe ring artifacts due to the response non-uniformity and non-linearity of detector units, which diminishes the ability to detect defects effectively.MethodsTo address this issue, we propose a physics-based ring artifacts reduction method for CT that employs pixel response correction. This method physically accounts for the cause of ring artifacts and leverages the prior knowledge of the detected object to enhance the accuracy of the detection process.ResultsOur proposed method achieved a notable reduction in ring artifacts, as evidenced by a 37.7% decrease in ring total variation (RTV) values compared to the originals, significantly enhancing image quality. It also surpassed traditional and machine learning methods in artifact reduction while maintaining image details. The lower RTV scores confirm our method's superior effectiveness in minimizing ring artifacts.ConclusionWe believe that our research contributes to the enhancement of defect inspection performance in detection systems, which is crucial for ensuring the safety of reactors. The proposed method's effectiveness in mitigating ring artifacts while maintaining image quality highlights its potential impact on the reliability of non-destructive testing in the context of HTGR components.

Background: Alzheimer's disease (AD) is a neurodegenerative disorder. There are no drugs and methods for the treatment of AD, but early intervention can delay the deterioration of the disease. Therefore, the early diagnosis of AD and mild cognitive impairment (MCI) is significant. Structural magnetic resonance imaging (sMRI) is widely used to present structural changes in the subject's brain tissue. The relatively mild structural changes in the brain with MCI have led to ongoing challenges in the task of conversion prediction in MCI. Moreover, many multimodal AD diagnostic models proposed in recent years ignore the potential relationship between multimodal information.

Objective: To solve these problems, we propose a multimodal fine-grained classification model based on deep metric learning for AD diagnosis (DML-MFCM), which can fully exploit the fine-grained feature information of sMRI and learn the potential relationships between multimodal feature information.

Methods: First, we propose a fine-grained feature extraction module that can effectively capture the fine-grained feature information of the lesion area. Then, we introduce a multimodal cross-attention module to learn the potential relationships between multimodal data. In addition, we design a hybrid loss function based on deep metric learning. It can guide the model to learn the feature representation method between samples, which improves the model's performance in disease diagnosis.

Results: We have extensively evaluated the proposed models on the ADNI and AIBL datasets. The ACC of AD vs. NC, MCI vs. NC, and sMCI vs. pMCI tasks in the ADNI dataset are 98.75%, 95.88%, and 88.00%, respectively. The ACC on the AD vs. NC and MCI vs. NC tasks in the AIBL dataset are 94.33% and 91.67%.

Conclusions: The results demonstrate that our method has excellent performance in AD diagnosis.

Background: K-edge subtraction (KES) tomography has been extensively utilized in the field of elemental sensitive imaging due to its high spatial resolution, rapid acquisition, and three-dimensional (3D) imaging capabilities. However, previous studies have primarily focused on the qualitative analysis of element contents, rather than quantitative assessment.

Objective: The current study proposes a novel method for quantitative elemental analysis based on K-edge subtraction tomography.

Methods: The linear correlation between the slice grayscale of standard samples and the difference in their linear absorption coefficients is confirmed. This finding suggests that the grayscale data from slices may be employed to perform quantitative estimations of elemental compositions.

Results: In order to verify the accuracy and validity of this method, the target element contents in standard and actual samples are quantitatively analyzed, respectively. The results demonstrate that the method is capable of achieving nanometer-resolved quantitative elemental sensitive imaging with a relative error of less than 3% in the target elemental content.

Conclusions: The method described in this paper is expected to expand the scope of applications for K-edge subtraction tomography and provide a novel approach to achieve more precise and convenient quantitative elemental analysis.

Background: Osteoporosis (OP) is a condition that significantly decreases bone density and strength, often remaining undetected until the occurrence of a fracture. Timely identification of OP is essential for preventing fractures, reducing morbidity, and enhancing the quality of life.

Objective: This study aims to improve the accuracy, speed, and reliability of early-stage osteoporosis detection by integrating the compact architecture of Cascaded ShuffleNet with the pattern recognition prowess of Artificial Neural Networks (ANNs).

Methods: BoneVoyage leverages the efficiency of ShuffleNet and the analytical capabilities of ANNs to extract and analyze complex features from bone density scans. The framework was trained and validated on a comprehensive dataset containing thousands of bone density images, ensuring robustness across diverse cases.

Results: This model achieving an accuracy of 97.2%, with high sensitivity and specificity. These results significantly surpass those of existing OP detection methods, highlighting the effectiveness of the BoneVoyage framework in identifying subtle changes in bone density indicative of early-stage osteoporosis.

Conclusions: BoneVoyage represents a significant advancement in the early detection of osteoporosis, offering a reliable tool for healthcare providers to identify at-risk individuals prematurely. The early detection facilitated by BoneVoyage allows for the implementation of preventive measures and targeted treatments.

Due to the hardware constraints of Computed Tomography (CT) imaging, acquiring high-resolution (HR) CT images in clinical settings poses a significant challenge. In recent years, convolutional neural networks have shown great potential in CT super-resolution (SR) problems. However, the reconstruction results of many deep learning-based SR methods have structural distortion and detail ambiguity. In this paper, a new SR network based on generative adversarial learning is proposed. The network consists of gradient branch and SR branch. Gradient branch is used to recover HR gradient maps. The network merges gradient image features of the gradient branch into the SR branch, offering gradient information guidance for super-resolution (SR) reconstruction. Further, the loss function of the network combines the image space loss function with the gradient loss and the gradient variance loss to further generate a more realistic detail texture. Compared to other comparison algorithms, the structural similarity index of the SR results obtained by the proposed method on simulation and experimental data has increased by 1.8% and 1.4%, respectively. The experimental results demonstrate that the proposed CT SR network exhibits superior performance in terms of structure preservation and detail restoration.

Background: The development of automatic methods for vertebral segmentation provides the objective analysis of each vertebra in the spine image, which is important for the diagnosis of various spinal diseases. However, vertebrae have inter-class similarity and intra-class variability, and some adjacent vertebrae exhibit adhesion.

Objective: To solve the adhesion problem of adjacent vertebrae and ensure that the boundary between adjacent vertebrae can be accurately demarcated, we propose an image segmentation method based on deep learning and marker controlled watershed.

Methods: This method consists of a dual-path model of localization path and segmentation path to achieve automatic vertebral segmentation. For the vertebral localization path, a high-resolution network (HRNet) is used to locate vertebral center. Moreover, based on spine posture, a new bone direction loss (BD-Loss) is designed to constrain HRNet. For the vertebral segmentation path, we proposed a VU-Net network to achieve vertebral preliminary segmentation. Additionally, a position information perception module (PIPM) is introduced to realize the guidance of HRNet to VU-Net. Finally, we novelly use the outputs of HR-Net and VU-Net deep learning networks to initialize the marker controlled watershed algorithm to suppress the adhesion of adjacent vertebrae and achieve vertebral fine segmentation.

Results: The proposed method was evaluated on two spine X-ray datasets using four metrics. The first dataset contains sagittal images of the cervical spine, while the second dataset contains coronal images of the whole spine, both with different health conditions. Our method achieved Recall of 96.82% and 94.38%, Precision of 97.24% and 98.14%, Dice coefficient of 97.03% and 96.22%, Intersection over Union of 94.24% and 92.72% on the cervical spine and whole spine datasets respectively, outperforming current state-of-the-art techniques.

Background and objective: This study aims to assess the effectiveness of combining radiomics features (RFs) with deep learning features (DFs) for classifying brain tumors-specifically Glioma, Meningioma, and Pituitary Tumor-using MRI scans and advanced ensemble learning techniques.

Methods: A total of 3064 T1-weighted contrast-enhanced brain MRI scans were analyzed. RFs were extracted using Pyradiomics, while DFs were obtained from a 3D convolutional neural network (CNN). These features were used both individually and together to train a range of machine learning models, including Support Vector Machines (SVM), Decision Trees (DT), Random Forests (RF), AdaBoost, Bagging, k-Nearest Neighbors (KNN), and Multi-Layer Perceptrons (MLP). To enhance the accuracy of these models, ensemble approaches such as Stacking, Voting, and Boosting were employed. LASSO feature selection and five-fold cross-validation were utilized to ensure the models' robustness.

Results: The results demonstrated that combining RFs and DFs significantly improved the model's performance compared to using either feature set alone. The best performance was achieved using the combined RF + DF approach with ensemble methods, particularly Boosting, which resulted in an accuracy of 95.0%, an AUC of 0.92, a sensitivity of 88%, and a specificity of 90%. Conversely, models utilizing only RFs or DFs showed lower performance, with RFs reaching an AUC of 0.82 and DFs achieving an AUC of 0.85.

Conclusion: The integration of RFs and DFs, along with advanced ensemble methods, significantly improves the accuracy and reliability of brain tumor classification using MRI. This approach shows strong clinical potential, with opportunities for further enhancing generalizability and precision through additional MRI sequences and advanced machine learning techniques.

Purpose: The periodic quality control of X-ray devices is important for obtaining optical medical images and determining the appropriate X-ray exposure dose. Additionally, the measurement of the X-ray output is constrained by time, technical aspects, and expenses. Therefore, we investigated the usefulness of a simple method for managing X-ray output using an Exposure Index (EI).

Methods: The entire surface of the flat panel detector was X-ray-irradiated every Friday at the time of end-of-work inspection under the condition that the recorded EI was approximately 1000. The EI and exposure dose were measured, and the linearity and accuracy were evaluated.

Results: The output gradually decreased from the start of the measurements in Room 1 and stabilized after the output was adjusted. The relationship between exposure dose and EI showed high linearity, with R² > 0.99, and the CV of EI was less than 2.41%, indicating high reproducibility.

Conclusions: We demonstrated that the results of constancy tests can be easily quantified using EI. The EI method can manage the X-ray output with good reproducibility.

Background: Computed tomography angiography (CTA) provides significant information on image quality in vascular imaging, thus offering high-resolution images despite having the disadvantages of increased radiation doses and contrast agent-related side effects. The deep-learning image reconstruction strategies were used to quantitatively evaluate the enhanced contrast-to-noise ratio (CNR) and the dose reduction effect of subtracted images.

Objective: This study aimed to elucidate a comprehensive understanding of the quantitative image quality features of the conventional filtered back projection (FBP) and the advanced intelligent clear-IQ engine (AiCE), a deep learning reconstruction technique. The comparison was made in subtracted images with variable concentrations of contrast agents at variable tube currents and voltages, enhancing our knowledge of these two techniques.

Methods: Data were obtained using a state-of-the-art 320-detector CT scanner. Image reconstruction was performed using FBP and AiCE with various intensities. The image quality evaluation was based on eight iodine concentrations in the phantom setup. The efficiency of AiCE relative to FBP was assessed by computing parameters including the root mean square error (RMSE), dose-dependent CNR, and potential dose reduction.

Results: The results showed that elevated concentrations of iodine and increased tube currents improved AiCE performance regarding CNR enhancement compared to FBP. AiCE also demonstrated a potential dose reduction ranging from 13.7 to 81.9% compared to FBP, suggesting a significant reduction in radiation exposure while maintaining image quality.

Conclusions: The employment of deep learning image reconstruction with AiCE presented a significant improvement in CNR and potential dose reduction in CT angiography. This study highlights the potential of AiCE to improve vascular image quality and decrease radiation exposure risk, thereby improving diagnostic precision and patient care in vascular imaging practices.

Background: Nuclear graphite and carbon components are vital structural elements in the cores of high-temperature gas-cooled reactors(HTGR), serving crucial roles in neutron reflection, moderation, and insulation. The structural integrity and stable operation of these reactors heavily depend on the quality of these components. Helical Computed Tomography (CT) technology provides a method for detecting and intelligently identifying defects within these structures. However, the scarcity of defect datasets limits the performance of deep learning-based detection algorithms due to small sample sizes and class imbalance.

Objective: Given the limited number of actual CT reconstruction images of components and the sparse distribution of defects, this study aims to address the challenges of small sample sizes and class imbalance in defect detection model training by generating approximate CT reconstruction images to augment the defect detection training dataset.

Methods: We propose a novel CT detection image generation algorithm called the Decompound Synthesize Method (DSM), which decomposes the image generation process into three steps: model conversion, background generation, and defect synthesis. First, STL files of various industrial components are converted into voxel data, which undergo forward projection and image reconstruction to obtain corresponding CT images. Next, the Contour-CycleGAN model is employed to generate synthetic images that closely resemble actual CT images. Finally, defects are randomly sampled from an existing defect library and added to the images using the Copy-Adjust-Paste (CAP) method. These steps significantly expand the training dataset with images that closely mimic actual CT reconstructions.

Results: Experimental results validate the effectiveness of the proposed image generation method in defect detection tasks. Datasets generated using DSM exhibit greater similarity to actual CT images, and when combined with original data for training, these datasets enhance defect detection accuracy compared to using only the original images.

Conclusion: The DSM shows promise in addressing the challenges of small sample sizes and class imbalance. Future research can focus on further optimizing the generation algorithm and refining the model structure to enhance the performance and accuracy of defect detection models.