3D Computed Tomography Images Research Articles

Image-based microstructural simulation of thermal conductivity for highly porous wood fiber insulation boards

The thermal conductivity of wood fiber insulation boards is significantly influenced by the microstructure of the fiber network and in general, the efficiency of wood fiber insulation boards increases with porosity. For higher raw densities, the raw density is a good predictor for the thermal conductivity. For lower raw densities however, this simple relation does not hold anymore. Here, structural information gained from 3D computed tomography images at several scales, modeling of the microstructure, and numerical simulation of the thermal conductivity are combined to get deeper insight into which and how microstructural features influence the thermal conductivity. The model-based simulation as described here shows that the presence and orientation of wood fiber clusters impact the thermal conductivity significantly.

Introduction of a bone-centered three-dimensional coordinate system enables computed tomographic canine femoral angle measurements independent of positioning.

Measurement of torsional deformities and varus alignment in the canine femur is clinically and surgically important but difficult. Computed tomography (CT) generates true three-dimensional (3D) information and is used to overcome the limitations of radiography. The 3D CT images can be rotated freely, but the final view for angle measurements remains a subjective variable decision, especially in severe and complex angular and torsional deformities. The aim of this study was the development of a technique to measure femoral angles in a truly three-dimensional way, independent of femoral positioning. To be able to set reference points in any image and at arbitrary positions of the CT series, the 3D coordinates of the reference points were used for mathematical calculation of the angle measurements using the 3D medical imaging Software VoXim®. Anatomical reference points were described in multiplanar reconstructions and volume rendering CT. A 3D bone-centered coordinate system was introduced and aligned with the anatomical planes of the femur. For torsion angle measurements, the transverse projection plane was mathematically defined by orthogonality to the longitudinal diaphyseal axis. For varus angle measurements, the dorsal plane was defined by a femoral retrocondylar axis. Independence positioning was tested by comparison of angle measurement results in repeated scans of 13 femur bones in different parallel and two double oblique (15/45°) positions in the gantry. Femoralvarus (or valgus), neck version (torsion), and inclination angles were measured, each in two variations. Resulting mean differences ranged between -0.9° and 1.3° for all six determined types of angles and in a difference of <1° for 17 out of 18 comparisons by subtraction of the mean angles between different positions, with one outlier of 1.3°. Intra- and inter-observer agreements determined by repeated measurements resulted in coefficients of variation for repeated measurements between 0.2 and 13.5%. The introduction of a bone-centered 3D coordinate system and mathematical definition of projection planes enabled 3D CT measurements of canine femoral varus and neck version and inclination angles. Agreement between angular measurements results of bones scanned in different positions on the CT table demonstrated that the technique is independent of femoral positioning.

Fast algorithms based on Empirical Interpolation Methods for selecting best projections in Sparse-View X-ray Computed Tomography using a priori information

To ensure the safety of critical parts in mass production, non-destructive techniques are increasingly considered. Industrial Computed Tomography (CT) is a unique tool to inspect a part integrity and dimensional conformity. However, it usually requires many radiographies (also called projections) to reconstruct a 3D CT image accurately. To reconstruct an object from a reduced number of projections has become one of the most crucial challenges in X-ray CT. Iterative reconstruction algorithms have relieved robotic CT systems from usual circular or helical trajectories, allowing them to focus on the most informative and relevant projections. In this paper, we propose an approach to select the best projections for X-ray tomographic reconstructions when a priori information on the inspected object is available. The presented algorithms are based on Empirical Interpolation Methods and show a good reconstruction quality from a dozen of projections. Those methods have been selected for their flexibility, speed, and the perspectives they open up for sinogram inpainting.

3D CNN-Residual Neural Network Based Multimodal Medical Image Classification

Multimodal medical imaging has become incredibly common in the area of biomedical imaging. Medical image classification has been used to extract useful data from multimodality medical image data. Magnetic resonance imaging (MRI) and Computed tomography (CT) are some of the imaging methods. Different imaging technologies provide different imaging information for the same part. Traditional ways of illness classification are effective, but in today's environment, 3D images are used to identify diseases. In comparison to 1D and 2D images, 3D images have a very clear vision. The proposed method uses 3D Residual Convolutional Neural Network (CNN ResNet) for the 3D image classification. Various methods are available for classifying the disease, like cluster, KNN, and ANN. Traditional techniques are not trained to classify 3D images, so an advanced approach is introduced in the proposed method to predict the 3D images. Initially, the multimodal 2D medical image data is taken. This 2D input image is turned into 3D image data because 3D images give more information than the 2D image data. Then the 3D CT and MRI images are fused and using the Guided filtering, and the combined image is filtered for the further process. The fused image is then augmented. Finally, this fused image is fed to 3DCNN ResNet for classification purposes. The 3DCNN ResNet classifies the image data and produces the output as five different stages of the disease. The proposed method achieves 98% of accuracy. Thus the designed modal has predicted the stage of the disease in an effective manner.

3D renal model for surgical planning of partial nephrectomy: A way to improve surgical outcomes.

to evaluate the impact of 3D model for a comprehensive assessment of surgical planning and quality of partial nephrectomy (PN). 195 patients with cT1-T2 renal mass scheduled for PN were enrolled in two groups: Study Group (n= 100), including patients referred to PN with revision of both 2D computed tomography (CT) imaging and 3D model; Control group (n= 95), including patients referred to PN with revision of 2D CT imaging. Overall, 20 individuals were switched to radical nephrectomy (RN). The primary outcome was the impact of 3D models-based surgical planning on Trifecta achievement (defined as the contemporary absence of positive surgical margin, major complications and ≤30% postoperative eGFR reduction). The secondary outcome was the impact of 3D models on surgical planning of PN. Multivariate logistic regressions were used to identify predictors of selective clamping and Trifecta's achievement in patients treated with PN (n=175). Overall, 73 (80.2%) patients in Study group and 53 (63.1%) patients in Control group achieved the Trifecta (p=0.01). The preoperative plan of arterial clamping was recorded as clampless, main artery and selective in 22 (24.2%), 22 (24.2%) and 47 (51.6%) cases in Study group vs. 31 (36.9%), 46 (54.8%) and 7 (8.3%) cases in Control group, respectively (p<0.001). At multivariate logistic regressions, the use of 3D model was found to be independent predictor of both selective or super-selective clamping and Trifecta's achievement. 3D-guided approach to PN increase the adoption of selective clamping and better predict the achievement of Trifecta.

Multiscale Pore Structure Evolution of Longmaxi Shale Induced by Acid Treatment

SummaryHydraulic fracturing to generate complex fracture networks is essential for shale reservoir development. However, the recovery of shale oil and gas is still low due to various engineering and geological factors. Acid treatment has been approved as a potential approach to enhance stimulated reservoir volume (SRV) by changing petrophysical and mechanical properties. Understanding the multiscale pore structure evolution behind the macro-performance change is critical in the application of acid treatment in shale reservoirs. In this study, cylindrical and powder shale samples from the Longmaxi formation are treated with 15 wt% hydrochloric acid (HCl) for 10 days. Before and after acid treatment, X-ray computed tomography (CT) and N2 adsorption techniques are used to characterize shale pore structure at microscale and nanoscale, respectively. Combined with the determination of variations in chemical compositions of shale samples and acid solutions, the mechanism of multiscale pore structure evolution induced by acid treatment is discussed. The N2 adsorption results uncover a considerable increase in volume and size of nanopores. All the nanopores increase in carbonate-rich shale, whereas the micropores and mesopores undergo a decrease in clay-rich shale. Reconstructed 3D CT images reveal the generation of large volumes of microscale pores and fractures, which leads to an increase in porosity of about 9%. The pore structure evolution in shale due to acid treatment is controlled by both mineralogy and microstructure. These findings demonstrate the promise of acid treatment for enhanced SRV and long-term productivity of shale oil and gas reservoirs in China.

Investigating the bone regeneration activity of PVA nanofibers scaffolds loaded with simvastatin/chitosan nanoparticles in an induced bone defect rabbit model

This study aims at preparing electrospun PVA NFs incorporating simvastatin/chitosan nanoparticles (SIM CS NPs) as a controlled drug eluting scaffold for bone regeneration. Optimization was performed by Design Expert® software through establishing two factor, three level factorial design, where the independent variables were the applied voltage, flow rate and PVA solution/SIM CS NPs ratio. Formulation variables values for the optimized formula were 18KV, 0.5 mL/h, and 3:1 respectively. NFs diameter and mesh pore size were chosen as the dependent variables. The optimized NFs were evaluated morphologically, chemically, and physically. Additionally, in-vitro SIM release from the scaffolds was investigated along 24 days. Optimum NFs possessed 136 nm diameter size and 6.5 nm porosity. Also, they showed sustained SIM release for 24 days to achieve the desired goal in bone regeneration. The optimized NFs were implanted within induced bone defects in rabbits. In-vivo assessments were performed through cone beam computed tomography 3D images, bone density measurements, histological analysis and bone morphogenetic protein 2 (BMP 2) level. The obtained results proved the high potential of the optimized NFs in promoting bone regeneration compared to untreated group, non-medicated NFs group, free SIM group and NFs loaded with SIM group after 6 weeks of implantation.

Segmentation of liver tumors with abdominal computed tomography using fully convolutional networks.

Dividing liver organs or lesions depicting on computed tomography (CT) images could be applied to help tumor staging and treatment. However, most existing image segmentation technologies use manual or semi-automatic analysis, making the analysis process costly and time-consuming. This research aims to develop and apply a deep learning network architecture to segment liver tumors automatically after fine tuning parameters. The medical imaging is obtained from the International Symposium on Biomedical Imaging (ISBI), which includes 3D abdominal CT scans of 131 patients diagnosed with liver tumors. From these CT scans, there are 7,190 2D CT images along with the labeled binary images. The labeled binary images are regarded as gold standard for evaluation of the segmented results by FCN (Fully Convolutional Network). The backbones of FCN are extracted from Xception, InceptionresNetv2, MobileNetv2, ResNet18, ResNet50 in this study. Meanwhile, the parameters including optimizers (SGDM and ADAM), size of epoch, and size of batch are investigated. CT images are randomly divided into training and testing sets using a ratio of 9:1. Several evaluation indices including Global Accuracy, Mean Accuracy, Mean IoU (Intersection over Union), Weighted IoU and Mean BF Score are applied to evaluate tumor segmentation results in the testing images. The Global Accuracy, Mean Accuracy, Mean IoU, Weighted IoU, and Mean BF Scores are 0.999, 0.969, 0.954, 0.998, 0.962 using ResNet50 in FCN with optimizer SGDM, batch size 12, and epoch 9. It is important to fine tuning the parameters in FCN model. Top 20 FNC models enable to achieve higher tumor segmentation accuracy with Mean IoU over 0.900. The occurred frequency of InceptionresNetv2, MobileNetv2, ResNet18, ResNet50, and Xception are 9, 6, 3, 5, and 2 times. Therefore, the InceptionresNetv2 has higher performance than others. This study develop and test an automated liver tumor segmentation model based on FCN. Study results demonstrate that many deep learning models including InceptionresNetv2, MobileNetv2, ResNet18, ResNet50, and Xception have high potential to segment liver tumors from CT images with accuracy exceeding 90%. However, it is still difficult to accurately segment tiny and small size tumors by FCN models.

Deterministic small-scale undulations of image-based risk predictions from the deep learning of lung tumors in motion.

Deep learning (DL) models that use medical images to predict clinical outcomes are poised for clinical translation. For tumors that reside in organs that move, however, the impact of motion (i.e., degenerated object appearance or blur) on DL model accuracy remains unclear. We examine the impact of tumor motion on an image-based DL framework that predicts local failure risk after lung stereotactic body radiotherapy (SBRT). We input pre-therapy free breathing (FB) computed tomography (CT) images from 849 patients treated with lung SBRT into a multitask deep neural network to generate an image fingerprint signature (or DL score) that predicts time-to-event local failure outcomes. The network includes a convolutional neural network encoder for extracting imaging features and building a task-specific fingerprint, a decoder for estimating handcrafted radiomic features, and a task-specific network for generating image signature for radiotherapy outcome prediction. The impact of tumor motion on the DL scores was then examined for a holdout set of 468 images from 39 patients comprising: (1) FB CT, (2) four-dimensional (4D) CT, and (3) maximum-intensity projection (MIP) images. Tumor motion was estimated using a 3D vector of the maximum distance traveled, and its association with DL score variance was assessed by linear regression. The variance and amplitude in 4D CT image-derived DL scores were associated with tumor motion (R2 =0.48 and 0.46, respectively). Specifically, DL score variance was deterministic and represented by sinusoidal undulations in phase with the respiratory cycle. DL scores, but not tumor volumes, peaked near end-exhalation. The mean of the scores derived from 4D CT images and the score obtained from FB CT images were highly associated (Pearson r=0.99). MIP-derived DL scores were significantly higher than 4D- or FB-derived risk scores (p<0.0001). An image-based DL risk score derived from a series of 4D CT images varies in a deterministic, sinusoidal trajectory in a phase with the respiratory cycle. These results indicate that DL models of tumors in motion can be robust to fluctuations in object appearance due to movement and can guide standardization processes in the clinical translation of DL models for patients with lung cancer.

Transformer-based factorized encoder for classification of pneumoconiosis on 3D CT images

In the past decade, deep learning methods have been implemented in the medical image fields and have achieved good performance. Recently, deep learning algorithms have been successful in the evaluation of diagnosis on lung images. Although chest radiography (CR) is the standard data modality for diagnosing pneumoconiosis, computed tomography (CT) typically provides more details of the lesions in the lung. Thus, a transformer-based factorized encoder (TBFE) was proposed and first applied for the classification of pneumoconiosis depicted on 3D CT images. Specifically, a factorized encoder consists of two transformer encoders. The first transformer encoder enables the interaction of intra-slice by encoding feature maps from the same slice of CT. The second transformer encoder explores the inter-slice interaction by encoding feature maps from different slices. In addition, the lack of grading standards on CT for labeling the pneumoconiosis lesions. Thus, an acknowledged CR-based grading system was applied to mark the corresponding pneumoconiosis CT stage. Then, we pre-trained the 3D convolutional autoencoder on the public LIDC-IDRI dataset and fixed the parameters of the last convolutional layer of the encoder to extract CT feature maps with underlying spatial structural information from our 3D CT dataset. Experimental results demonstrated the superiority of the TBFE over other 3D-CNN networks, achieving an accuracy of 97.06%, a recall of 89.33%, precision of 90%, and an F1-score of 93.33%, using 10-fold cross-validation.

Development and Validation of a Mobile Application for Measuring Tibial Torsion.

Tibial torsion lacks a single and reliable method for its measurement. While physical examination, computed tomography (CT), and EOS imaging are used complementarily, three-dimensional (3D) CT is the most widely used method for intuitive documentation and visualization. However, concern regarding the associated radiation hazard limits its use in the evaluation of pediatric patients. Moreover, EOS machines are too expensive and too large to be placed in every clinic requiring the measurement of tibial torsion. Therefore, a new method for 3D reconstruction is needed. In the present study, we tested the validity and reliability of a novel reconstruction tool for the lower leg. A statistical shape model and Laplacian constraint were adopted for the development of a new reconstruction tool for measuring tibial torsion. Tibial torsion measurements based on a 3D reconstruction application and CT images for 36 patients were evaluated for intraobserver and interobserver reliability. Tibial torsion measurements for 75 patients were compared for validation. A 3D reconstruction system for the lower leg was developed as a mobile application and was installed in a portable device for easy access in the clinical setting. In terms of interobserver reliability, the intraclass correlation coefficient among 3 clinicians was 0.896 (95% confidence interval [CI], 0.828 to 0.941). The correlation coefficient between tibial torsion measured with use of 3D CT and that measured with the mobile application was 0.865 (p < 0.001). The mobile application showed excellent reliability and validity for measuring tibial torsion. Concurrent utilization with mobile application for the femur allows visualization of the rotational profile of the leg without the need for CT or EOS. Diagnostic Level III . See Instructions for Authors for a complete description of levels of evidence.

Anatomical Characteristics of the Accessory Maxillary Ostium in Three-Dimensional Analysis.

Background and Objectives: The accessory maxillary ostium (AMO) can interfere with ventilation and drainage of the maxillary sinus, and therefore the importance of evaluating the anatomical features of the AMO has been emphasized. This study aimed to evaluate anatomical characteristics of the AMO together with the natural ostium (NO) using three-dimensional (3D) analysis and to assess the relationship between the AMO and maxillary sinus pathologies. Materials and Methods: This retrospective study included 394 sinuses in 197 patients. Using 3D computed tomography images, the prevalence of the AMO and concurrent sinus pathologies were examined. For patients with an AMO, 3D spatial positions of the AMO and NO related to adjacent anatomic structures and dimensions of the AMO and NO were evaluated. Results: A total of 84 sinuses showed single or multiple AMO, with a prevalence of 21.3%. The AMO was located superiorly by 30.1 mm from the maxillary sinus floor, inferiorly by 1.3 mm from the orbital floor, and posteriorly by 22.4 mm from the anterior sinus wall. The AMO was located 5.4 mm posteriorly and 0.7 mm inferiorly from the NO. On the same coronal plane as the NO or AMO, height from the maxillary sinus floor to the NO and AMO ranged from 19.4 to 45.8 mm and 14.5 mm to 41.9 mm, respectively. The mean horizontal and vertical dimensions were 5.9 mm and 4.6 mm for the NO and 2.8 mm and 3.0 mm for the AMO. We detected a significant association between the presence of the AMO and the mucosal thickening (p = 0.029). Conclusions: The results of this study suggest that, although the AMO and NO are mostly located in positions that do not limit sinus-related surgeries, such as maxillary sinus floor augmentation, the AMO and NO are also found in lower positions, which may be a detriment to the postoperative physiological function of the maxillary sinus and affect treatment outcomes.

CONTRIBUTION OF ORGAN-BASED TUBE CURRENT MODULATION TO THE REDUCTION OF LENS EXPOSURE DOSE IN HEAD 4D CT IMAGING: A PHANTOM STUDY.

The purpose of this study was to evaluate the effectiveness of organ effect modulation (OEM) in reducing the lens dose in 4D computed tomography (CT) of the head in volume-acquisition (NVA) mode. Six radiophotoluminescent dosemeters were placed on the head of a RANDO phantom. The doses absorbed by the organs and image noise change rate were determined. The lens doses without OEM (i.e. in the OEMoff case) were higher than those with the same target standard deviation and volume-computed tomography dose index (CTDIvol) as in the OEMoff case (p<0.01). The image noise change rate was 11%. OEM reduced the lens dose during head 4D CT imaging in the NVA mode by 18%. Furthermore, the feasibility of lens dose reduction while ensuring sufficient image quality was confirmed under the condition in which OEM was employed with the same CTDIvol as in the OEMoff case.

EP05.01-019 4D CT Ventilation Image-Guided Lung Functional Avoidance Radiotherapy: A Single-Arm Prospective Pilot Clinical Trial

Thoracic radiotherapy is limited by substantial toxicities, including radiation pneumonitis which is potentially fatal. Functional avoidance radiotherapy that preferentially avoids irradiating normal (ventilated or perfused) lung regions has shown potential to reduce pneumonitis. The primary objective of this single-arm prospective pilot clinical trial (NCT02308709) was to assess the safety and feasibility of functional avoidance radiotherapy with 4-dimensional (4D) computed tomography (CT) ventilation imaging, which only requires a standard-of-care 4D CT scan and image processing/analysis.

Effective lung ventilation estimation based on 4D CT image registration and supervoxels

Most computed tomography (CT)-derived ventilation estimation methods rely on deformation fields of image registration to track the volume variations of each voxel, which are easily affected by image registration results and motion artifacts of four-dimensional (4D) CT images. To attack these problems, a lung ventilation estimation method based on 4D CT image registration and supervoxels is proposed in this paper. In order to avoid estimation errors caused by image artifacts and noises in the methods of directly estimating voxel-wise ventilation, images corresponding to maximum exhale phase are represented by multi-level supervoxels. Then, according to the relationship between registered CT values in Hounsfield units (HU) and regional volume change, a ventilation estimation method is designed to calculate the whole ventilation of each supervoxel. To accurately recover the ventilation of each voxel from the supervoxel region, a linear programming model is established to obtain the ventilation of each voxel, and the final estimated lung ventilation image is obtained by averaging the recovered results of all supervoxels layers. To evaluate our proposed ventilation estimation method, we calculate voxel-wise Spearman coefficient rs and Dice similarity coefficient (DSC for low function lung DSClow and high function lung DSChigh) of various CT-derived ventilation methods on VAMPIRE dataset. The experimental results on VAMPIRE dataset show that the average rs values between our proposed method can achieve 0.38 (-0.17–0.70). Accordingly, the average DSClow and DSChigh values of our proposed method are 0.40 (0.17–0.67) and 0.36 (0.08–0.61) respectively.

On the Simulation of Ultra-Sparse-View and Ultra-Low-Dose Computed Tomography with Maximum a Posteriori Reconstruction Using a Progressive Flow-Based Deep Generative Model.

Ultra-sparse-view computed tomography (CT) algorithms can reduce radiation exposure for patients, but these algorithms lack an explicit cycle consistency loss minimization and an explicit log-likelihood maximization in testing. Here, we propose X2CT-FLOW for the maximum a posteriori (MAP) reconstruction of a three-dimensional (3D) chest CT image from a single or a few two-dimensional (2D) projection images using a progressive flow-based deep generative model, especially for ultra-low-dose protocols. The MAP reconstruction can simultaneously optimize the cycle consistency loss and the log-likelihood. We applied X2CT-FLOW for the reconstruction of 3D chest CT images from biplanar projection images without noise contamination (assuming a standard-dose protocol) and with strong noise contamination (assuming an ultra-low-dose protocol). We simulated an ultra-low-dose protocol. With the standard-dose protocol, our images reconstructed from 2D projected images and 3D ground-truth CT images showed good agreement in terms of structural similarity (SSIM, 0.7675 on average), peak signal-to-noise ratio (PSNR, 25.89 dB on average), mean absolute error (MAE, 0.02364 on average), and normalized root mean square error (NRMSE, 0.05731 on average). Moreover, with the ultra-low-dose protocol, our images reconstructed from 2D projected images and the 3D ground-truth CT images also showed good agreement in terms of SSIM (0.7008 on average), PSNR (23.58 dB on average), MAE (0.02991 on average), and NRMSE (0.07349 on average).

Double U-Net CycleGAN for 3D MR to CT image synthesis.

CycleGAN and its variants are widely used in medical image synthesis, which can use unpaired data for medical image synthesis. The most commonly used method is to use a Generative Adversarial Network (GAN) model to process 2D slices and thereafter concatenate all of these slices to 3D medical images. Nevertheless, these methods always bring about spatial inconsistencies in contiguous slices. We offer a new model based on the CycleGAN to work out this problem, which can achieve high-quality conversion from magnetic resonance (MR) to computed tomography (CT) images. To achieve spatial consistencies of 3D medical images and avoid the memory-heavy 3D convolutions, we reorganized the adjacent 3 slices into a 2.5D slice as the input image. Further, we propose a U-Net discriminator network to improve accuracy, which can perceive input objects locally and globally. Then, the model uses Content-Aware ReAssembly of Features (CARAFE) upsampling, which has a large field of view and content awareness takes the place of using a settled kernel for all samples. The mean absolute error (MAE), peak-signal-to-noise ratio (PSNR), and structural similarity index measure (SSIM) for double U-Net CycleGAN generated 3D image synthesis are 74.56±10.02, 27.12±0.71 and 0.84±0.03, respectively. Our method achieves preferable results than state-of-the-art methods. The experiment results indicate our method can realize the conversion of MR to CT images using ill-sorted pair data, and achieves preferable results than state-of-the-art methods. Compared with 3D CycleGAN, it can synthesize better 3D CT images with less computation and memory.

Robotic segmentectomy using a lung base-flip approach.

Robotic segmentectomy using a lung base-flip approach.

A wavelet-based deep learning pipeline for efficient COVID-19 diagnosis via CT slices

The quick diagnosis of the novel coronavirus (COVID-19) disease is vital to prevent its propagation and improve therapeutic outcomes. Computed tomography (CT) is believed to be an effective tool for diagnosing COVID-19, however, the CT scan contains hundreds of slices that are complex to be analyzed and could cause delays in diagnosis. Artificial intelligence (AI) especially deep learning (DL), could facilitate and speed up COVID-19 diagnosis from such scans. Several studies employed DL approaches based on 2D CT images from a single view, nevertheless, 3D multiview CT slices demonstrated an excellent ability to enhance the efficiency of COVID-19 diagnosis. The majority of DL-based studies utilized the spatial information of the original CT images to train their models, though, using spectral–temporal information could improve the detection of COVID-19. This article proposes a DL-based pipeline called CoviWavNet for the automatic diagnosis of COVID-19. CoviWavNet uses a 3D multiview dataset called OMNIAHCOV. Initially, it analyzes the CT slices using multilevel discrete wavelet decomposition (DWT) and then uses the heatmaps of the approximation levels to train three ResNet CNN models. These ResNets use the spectral–temporal information of such images to perform classification. Subsequently, it investigates whether the combination of spatial information with spectral–temporal information could improve the diagnostic accuracy of COVID-19. For this purpose, it extracts deep spectral–temporal features from such ResNets using transfer learning and integrates them with deep spatial features extracted from the same ResNets trained with the original CT slices. Then, it utilizes a feature selection step to reduce the dimension of such integrated features and use them as inputs to three support vector machine (SVM) classifiers. To further validate the performance of CoviWavNet, a publicly available benchmark dataset called SARS-COV-2-CT-Scan is employed. The results of CoviWavNet have demonstrated that using the spectral–temporal information of the DWT heatmap images to train the ResNets is superior to utilizing the spatial information of the original CT images. Furthermore, integrating deep spectral–temporal features with deep spatial features has enhanced the classification accuracy of the three SVM classifiers reaching a final accuracy of 99.33% and 99.7% for the OMNIAHCOV and SARS-COV-2-CT-Scan datasets respectively. These accuracies verify the outstanding performance of CoviWavNet compared to other related studies. Thus, CoviWavNet can help radiologists in the rapid and accurate diagnosis of COVID-19 diagnosis.

Calibration Method of Projectional Geometry for X-ray C-arm Fluoroscopy Using Sinogram Data

X-ray imaging represents the most commonly used imaging modality in X-ray-guided vascular intervention procedures. Computed tomography (CT) data ensure that the procedure is performed accurately and safely by providing medical staff with positional information of the body part before starting the procedure. In particular, accurate geometric information of the imaging equipment is essential to accurately calculate the three-dimensional (3D) position of catheters used in delicate operations. However, it is difficult to gather this information before surgery. Therefore, this study proposes a novel calibration method that can be used immediately before a procedure and can guarantee the stability of the procedure. The calibration was performed without additional radiography using sinogram data obtained during the 3D CT imaging process, and both the accuracy and calculation time available in the vascular intervention theater were allowable. The experimental results show that the best angular conditions in terms of calculations and accuracy are between −40 and 40 degrees in angular range and 1.6 degrees in angular interval. Consequently, we achieved a calculation time of 2.92 s and an average accuracy of 0.36 mm, thus meeting our goal of accuracy below 1 mm within a minute of computational time.