Resolution Computed Tomography Images Research Articles

Modularized data-driven reconstruction framework for nonideal focal spot effect elimination in computed tomography.

High-performance computed tomography (CT) plays a vital role in clinical decision-making. However, the performance of CT imaging is adversely affected by the nonideal focal spot size of the x-ray source or degraded by an enlarged focal spot size due to aging. In this work, we aim to develop a deep learning-based strategy to mitigate the problem so that high spatial resolution CT images can be obtained even in the case of a nonideal x-ray source. To reconstruct high-quality CT images from blurred sinograms via joint image and sinogram learning, a cross-domain hybrid model is formulated via deep learning into a modularized data-driven reconstruction (MDR) framework. The proposed MDR framework comprises several blocks, and all the blocks share the same network architecture and network parameters. In essence, each block utilizes two sub-models to generate an estimated blur kernel and a high-quality CT image simultaneously. In this way, our framework generates not only a final high-quality CT image but also a series of intermediate images with gradually improved anatomical details, enhancing the visual perception for clinicians through the dynamic process. We used simulated training datasets to train our model in an end-to-end manner and tested our model on both simulated and realistic experimental datasets. On the simulated testing datasets, our approach increases the information fidelity criterion (IFC) by up to 34.2%, the universal quality index (UQI) by up to 20.3%, the signal-to-noise (SNR) by up to 6.7%, and reduces the root mean square error (RMSE) by up to 10.5% as compared with FBP. Compared with the iterative deconvolution method (NSM), MDR increases IFC by up to 24.7%, UQI by up to 16.7%, SNR by up to 6.0%, and reduces RMSE by up to 9.4%. In the modulation transfer function (MTF) experiment, our method improves the MTF50% by 34.5% and MTF10% by 18.7% as compared with FBP, Similarly remarkably, our method improves MTF50% by 14.3% and MTF10% by 0.9% as compared with NSM. Also, our method shows better imaging results in the edge of bony structures and other tiny structures in the experiments using phantom consisting of ham and a bottle of peanuts. A modularized data-driven CT reconstruction framework is established to mitigate the blurring effect caused by a nonideal x-ray source with relatively large focal spot. The proposed method enables us to obtain high-resolution images with less ideal x-ray source.

Automatic segmentation of gross target volume of nasopharynx cancer using ensemble of multiscale deep neural networks with spatial attention

Radiotherapy is the main treatment method for nasopharynx cancer. Delineation of Gross Target Volume (GTV) from medical images is a prerequisite for radiotherapy. As manual delineation is time-consuming and laborious, automatic segmentation of GTV has a potential to improve the efficiency of this process. This work aims to automatically segment GTV of nasopharynx cancer from Computed Tomography (CT) images. However, it is challenged by the small target region, anisotropic resolution of clinical CT images, and the low contrast between the target region and surrounding soft tissues. To deal with these problems, we propose a 2.5D Convolutional Neural Network (CNN) to handle the different in-plane and through-plane resolutions. We also propose a spatial attention module to enable the network to focus on the small target, and use channel attention to further improve the segmentation performance. Moreover, we use a multi-scale sampling method for training so that the networks can learn features at different scales, which are combined with a multi-model ensemble method to improve the robustness of segmentation results. We also estimate the uncertainty of segmentation results based on our model ensemble, which is of great importance for indicating the reliability of automatic segmentation results for radiotherapy planning. Experiments with 2019 MICCAI StructSeg dataset showed that (1) Our proposed 2.5D network has a better performance on images with anisotropic resolution than the commonly used 3D networks. (2) Our attention mechanism can make the network pay more attention to the small GTV region and improve the segmentation accuracy. (3) The proposed multi-scale model ensemble achieves more robust results, and it can simultaneously obtain uncertainty information that can indicate potential mis-segmentations for better clinical decisions.

Characteristics of chest high resolution computed tomography images of COVID-19: A retrospective study of 46 patients.

Objective:To analyze the characteristics of chest high resolution computed tomography (CT) images of coronavirus disease 2019 (COVID-19).Methods:This is a retrospective study analyzing the clinical records and chest high-resolution CT images of 46 consecutive patients who were diagnosed with COVID-19 by nucleic acid tests and treated at our hospitals between January 2020 and February 2020.Results:Abnormalities in the CT images were found in 44 patients (95.6%). The lesions were unilateral in eight patients (17.4%), bilateral in 36 patients (78.3%), single in seven patients (15.9%), and multiple in 37 patients (84.1%). The morphology of the lesions was scattered opacity in 10 patients (21.7%), patchy opacity in 38 patients (82.6%), fibrotic cord in 17 patients (37.0%), and wedge-shaped opacity in two patients (4.3%). The lesions can be classified as ground-glass opacity in eight patients (17.4%), consolidation in one patient (2.2%), and ground-glass opacity plus consolidation in 28 patients (60.9%).Conclusion:Most COVID-19 patients showed abnormalities in chest CT images and the most common findings were ground-glass opacity plus consolidation.

A deep convolutional neural network for simultaneous denoising and deblurring in computed tomography

Computed tomography (CT) has been commonly used for providing medical images. However, CT scans have a potential risk of side-effect due to high radiation dose. In order to overcome this limitation, low-dose CT imaging techniques have been developed by using novel scanners, reconstruction algorithms and adjusted scanning protocols. In spite of these strategies, the conventional low-dose CT imaging techniques are hard to compromise between radiation dose and image quality, leading to noise, blur and artifacts in CT images. These drawbacks of the conventional techniques degrade diagnostic accuracy. In this study, we proposed a deep convolutional neural network (CNN)-based low-dose CT imaging technique for simultaneously reducing noise and blur at once. The proposed method, called denoising and deblurring CNN (DnDbCNN), consisted of two networks, and each network was separately trained for reducing noise and blur in CT images. For combining the trained networks, a convolutional layer was added between the two trained networks, and the combined network was re-trained to update the weights of each layer. Denoising convolutional neural network (DnCNN) and super-resolution convolutional neural network (SRCNN) were also implemented for comparison. We evaluated the performance of the DnDbCNN in terms of mean-square error (MSE), peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), coefficient-of-variation (COV) and modulation transfer function (MTF). The results showed that the DnDbCNN improved the noise property, spatial resolution and quantitative accuracy of CT images, and the performance of the DnDbCNN was superior to the conventional methods. As a result, the DnDbCNN can simultaneously reduce noise and blur by combining two networks through three training steps. Therefore, the DnDbCNN has a potential for improving CT image quality with low radiation dose and enables precise diagnosis.

Artificial Intelligence-related Literature in Transplantation: A Practical Guide.

Artificial Intelligence-related Literature in Transplantation: A Practical Guide.

Spectral Photon Counting CT: Imaging Algorithms and Performance Assessment.

Photon counting x-ray detectors (PCDs) with spectral capabilities have the potential to revolutionize computed tomography (CT) for medical imaging. The ideal PCD provides accurate energy information for each incident x-ray, and at high spatial resolution. This information enables material-specific imaging, enhanced radiation dose efficiency, and improved spatial resolution in CT images. In practice, PCDs are affected by non-idealities, including limited energy resolution, pulse pileup, and cross talk due to charge sharing, K-fluorescence, and Compton scattering. In order to maximize their performance, PCDs must be carefully designed to reduce these effects and then later account for them during correction and post-acquisition steps. This review article examines algorithms for using PCDs in spectral CT applications, including how non-idealities impact image quality. Performance assessment metrics that account for spatial resolution and noise such as the detective quantum efficiency (DQE) can be used to compare different PCD designs, as well as compare PCDs with conventional energy integrating detectors (EIDs). These methods play an important role in enhancing spectral CT images and assessing the overall performance of PCDs.

Relationship between the location of ligamentum flavum hypertrophy and its stress in finite element analysis

ObjectiveTo quantitatively describe the stress of the ligamentum flavum (LF) using the finite element method and to compare the stress at different parts of the healthy LF.MethodsBased on the high resolution computed tomography imaging data of a healthy 22‐year‐old man, three‐dimensional nonlinear L4–5 lumbar finite element model (FEM) representing intact condition was developed. The LF, as the object of the present research, was incorporated into the spinal model in the form of solid three‐dimensional structure. The model’s validity is verified by comparing its biomechanical indices, such as range of motion and axial compression pressure displacement, with published results under specific loading conditions. To authenticate the accuracy of the solid LF, the lamina attachments, the central cross‐section, and other anatomy indicators were compared with figures in the published literature. After the average and maximum von Mises stress on the surface of LF under various working conditions were measured using ANSYS and AutoCAD software, the surface stress difference in the LF between the ventral and dorsal sides as well as the lateral and lamina parts were determined.ResultsThe FEM predicted a similar tendency for biomechanical indices as shown in previous studies. The lamina attachments, the central cross‐section, and the height as well as the width of the LF in the healthy FEM were in accordance with published results. In the healthy model, the average and maximum von Mises stress in the shallow layer of the LF were, respectively, 1.40, 2.28, 1.76, 1.48, 1.38 and 1.79, 2.41, 1.46, 1.42, 1.71 times that in the deep layer under a compressive preload of 500 N incorporated with flexion, extension, and lateral and rotational moments (10 Nm). The most conspicuous difference in surface stress was observed with the flexion motion, with a nearly 241% difference in the maximum stress and a 228% difference in the average stress compared to those in other states. As far as the whole dorsal side of the LF was concerned, the maximum surface stress was almost all concentrated in the dorsal neighboring facet joint portion. In addition, the maximum and average stress were, respectively, 77%, 72%, 15%, 11%, 71% and 153%, 39%, 54%, 200%, 212% higher in the lateral part than in the lamina part.ConclusionBased on the predisposition of LF hypertrophy in the human spine and the stress distribution of this study, the positive correlation between LF hypertrophy and its stress was confirmed.

Coronavirus Disease 2019: Initial High Resolution Computed Tomography Imaging Feature Analysis: Report of 7 Cases.

The number of reported cases of the new coronavirus disease (COVID-19), caused by the severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2), has increased since December 2019. The initial high-resolution computed tomography (HRCT) images of 7 patients diagnosed with COVID-19 in the Affiliated Hospital of Hangzhou Normal University, China, were collected and analyzed. The study showed that all patients had had close contact with other COVID-19 patients and presented with fever. The initial white blood cell counts of all patients were normal. Subsequently, the percentage of lymphocytes decreased in 3 patients. In all 7 patients with COVID-19, ground-glass opacity (GGO) was found in the HRCT images, mainly distributed in the subpleural region of the lungs. The HRCT scans of 6 patients showed bilateral lobar lesions, with mainly peripheral subpleural distribution; 1 patient, instead, showed unilateral lobar involvement. The right lung was more extensively involved than the left lung in 6 patients, and the lower lobe was more extensively involved than the upper lobe in 5 patients. The initial chest HRCT images of the lungs of the analyzed COVID-19 patients had specific characteristics. The typical manifestation at both lungs was an extensive GGO-type infiltrate, with thickened vascular bundles and focal center consolidation. Pleural effusion, bilateral hilar, and mediastinal lymphadenopathy were rare.

Artificial Neural Network Implementation for Classification of Lung Tissues in High Resolution Tomography Images

In this digital world, the amount of information what we are getting is uncountable and speed at which it reaches us is unpredictable. We are getting is uncountable. In order to understand and predict the required data we should have the image classification technique otherwise there will be lot of confusion and misunderstanding will happen during its study. Content based image retrieval technique can be used to exploit the properties of high resolution computed tomography data which will be stored in the achieve by identifying the similar images which will helped for radiologist for self learning and differential diagnosis of intestinal lungs diseases. This technique identifies the particular image based on the content present in the image rather than keywords or tag related to the image. The content present in the image means, patterns, colors, textures etc. high resolution technique of image identification method is used in many medical applications like identifying diseases or the identification of symptoms of the diseases. in this project we are using the neural network classifier to classify the lung tissues in high resolution computed tomography images. This method gives the proper and clear classification of the lung tissues which will helps in medical science field. This neural network classifier uses the optimal subset feature to classify the pattern of the lung tissues.

Investigation of Sandstone Mesostructure Damage Caused by Freeze‐Thaw Cycles via CT Image Enhancement Technology

The mesostructures of rocks determine their macromechanical properties. These rock mesostructures may be altered by the freeze‐thaw cycles in cold regions. In this regard, this paper proposes a quantitative evaluation method based on computed tomography (CT) scanning technology for investigating the mesostructure and damage characteristics of sandstone subjected to freeze‐thaw conditions. CT scan images of two sandstones with different grain sizes were obtained after 0, 20, 40, 60, 80, and 100 freeze‐thaw cycles, using a high‐precision CT scanner. Based on the microphysical information contained in these CT images, pseudo‐color‐enhancement of the CT images of rocks subjected to freeze‐thaw cycles was realized. The use of such a pseudo‐color‐enhancement technique can improve the resolution of CT images. Thus, particle detachment, crack initiation, crack propagation, and increased porosity due to the volumetric expansion of water inside the rock could be detected and clearly observed. Furthermore, a numerical expression for the mesostructure and damage information contained in the pseudo‐color‐enhanced images is presented herein; this serves as a convenient method for quantitative analyses of sandstone damage under freeze‐thaw cycles. An analysis of the pseudo‐color‐enhanced images shows that, under freeze‐thaw cycles, damage propagation in sandstone originates from existing damage or defect sites. After the stages of crack (pore) formation, penetration, and propagation, the freeze‐thaw cycle‐induced damage increases gradually, while the effective bearing area of the rock decreases continuously. Herein, a schematic of a conceptual model for the freeze‐thaw cycle‐induced deterioration in sandstone mesostructures is presented. Damage propagation models for sandstones with two different grain sizes subjected to freeze‐thaw cycles were also developed. Based on the damage mechanics theory, a damage variable expressed in terms of the pore area was defined. Moreover, the relationship between this damage variable and the freeze‐thaw cycles was established.

Computationally efficient deep neural network for computed tomography image reconstruction.

Deep neural network-based image reconstruction has demonstrated promising performance in medical imaging for undersampled and low-dose scenarios. However, it requires large amount of memory and extensive time for the training. It is especially challenging to train the reconstruction networks for three-dimensional computed tomography (CT) because of the high resolution of CT images. The purpose of this work is to reduce the memory and time consumption of the training of the reconstruction networks for CT to make it practical for current hardware, while maintaining the quality of the reconstructed images. We unrolled the proximal gradient descent algorithm for iterative image reconstruction to finite iterations and replaced the terms related to the penalty function with trainable convolutional neural networks (CNN). The network was trained greedily iteration by iteration in the image domain on patches, which requires reasonable amount of memory and time on mainstream graphics processing unit (GPU). To overcome the local-minimum problem caused by greedy learning, we used deep UNet as the CNN and incorporated separable quadratic surrogate with ordered subsets for data fidelity, so that the solution could escape from easy local minimums and achieve better image quality. The proposed method achieved comparable image quality with state-of-the-art neural network for CT image reconstruction on two-dimensional (2D) sparse-view and limited-angle problems on the low-dose CT challenge dataset. The difference in root-mean-square-error (RMSE) and structural similarity index (SSIM) was within [-0.23,0.47]HU and [0,0.001], respectively, with 95% confidence level. For three-dimensional (3D) image reconstruction with ordinary-size CT volume, the proposed method only needed 2GB graphics processing unit (GPU) memory and 0.45s per training iteration as minimum requirement, whereas existing methods may require 417GB and 31min. The proposed method achieved improved performance compared to total variation- and dictionary learning-based iterative reconstruction for both 2D and 3D problems. We proposed a training-time computationally efficient neural network for CT image reconstruction. The proposed method achieved comparable image quality with state-of-the-art neural network for CT reconstruction, with significantly reduced memory and time requirement during training. The proposed method is applicable to 3D image reconstruction problems such as cone-beam CT and tomosynthesis on mainstream GPUs.

CT-image of rock samples super resolution using 3D convolutional neural network

Computed Tomography (CT) imaging technique is widely used in geological exploration, medical diagnosis and other fields. However, the resolution of CT images is usually limited by scanning devices and cost. Super-resolution (SR) methods based on deep learning provide remarkable performance for two-dimensional (2D) images. Unfortunately, few effective SR algorithms are available for three-dimensional (3D) images. This study proposes a novel network named as three-dimensional super-resolution convolutional neural network (3DSRCNN) to realize voxel SR imaging of rock samples. To solve the practical problems faced in the training process, such as slow convergence of network training and insufficient memory, we utilized adjustable learning rate, residual-learning, gradient clipping, momentum stochastic gradient descent (SGD) strategy to optimize training procedure. In addition, we have explored the empirical guidelines to set an appropriate number of network layers. Previous learning-based algorithms need to separately train samples for different scale factors; by contrast, our single model can perform the multi-scale SR. Further, our proposed method provides better performance in terms of PSNR, SSIM and efficiency compared with conventional methods.

CCD and scientific-CMOS detectors for submicron laboratory based X-ray Computed tomography

The key component of any CT (X-ray computed tomography) machine is a detection system. In area of scientific CT imaging applications three types of sensors are mainly used. These are amorphous silicon (α-Si) flat panels, complementary metal–oxide–semiconductor (CMOS) and charge-coupled device (CCD) sensors. Here the performance and comparison study of the two lastly named sensor types is conducted in field of high resolution CT imaging. CCD and novel sCMOS-based (scientific CMOS) detection systems are tested using high-resolution laboratory-based Rigaku nano3DX system (ability to achieve submicron voxel resolution). Properties of each camera were evaluated as well as the quality and noise properties of acquired data (both projection and CT data).

CT IMAGE RECONSTRUCTION ON A LOW DIMENSIONAL MANIFOLD.

The patch manifold of a natural image has a low dimensional structure and accommodates rich structural information. Inspired by the recent work of the low-dimensional manifold model (LDMM), we apply the LDMM for regularizing X-ray computed tomography (CT) image reconstruction. This proposed method recovers detailed structural information of images, significantly enhancing spatial and contrast resolution of CT images. Both numerically simulated data and clinically experimental data are used to evaluate the proposed method. The comparative studies are also performed over the simultaneous algebraic reconstruction technique (SART) incorporated the total variation (TV) regularization to demonstrate the merits of the proposed method. Results indicate that the LDMM-based method enables a more accurate image reconstruction with high fidelity and contrast resolution.

Automatic Knot Detection in Coarse-Resolution Cone-Beam Computed Tomography Images of Softwood Logs

Abstract X-ray computed tomography (CT) scanning of sawmill logs is associated with costly and complex machines. An alternative scanning solution was developed, but its data have not been evaluated regarding detection of internal features. In this exploratory study, a knot detection algorithm was applied to images of four logs to evaluate its performance in terms of knot position and size. The results were a detection rate of 67 percent, accurate position, and inaccurate size. Although the sample size was small, it was concluded that automatic knot detection in coarse resolution CT images of softwoods is feasible, albeit for knots of sufficient size.

Use Stereoscopic Model in Interventional and Surgical Procedures.

The 3-dimensional (3D) stereoscopic modeling software allows anatomists to create high-resolution 3D models from computed tomography (CT) images. In this paper, we used high resolution CT images from a cadaver and a patient to develop clinically relevant anatomic models that can be used to teach surgical trainees different surgical procedures and approaches. The model facilitates visualization, manipulation, and interaction. It can be presented in stereoscopic 3D in a virtual environment, either in a classroom setting or immediately preceding a surgical procedure.

High Resolution Computed Tomography and Magnetic Resonance Imaging in the Preoperative Assessment for Cochlear Implantation Candidacy

AbstractBackground: High Resolution Computed Tomography (HRCT) and magnetic resonance imaging are regularly used for cochlear implant preoperative evaluation for the evaluation of inner ear malformations, surgical planning, and especially the imaging of the VIIIth nerve. In children, these imaging procedures are especially important due to the high incidence of inner ear malformations.Aim of Work: To study the role of high resolution com-puted tomography and magnetic resonance imaging in clas-sifying 70 patients with profound deafness for possible absolute contraindications for surgery, relative contraindica-tions for surgery (difficult surgery) or easy surgery for cochlear implant.Subjects and Methods: We performed a cross sectional descriptive study on 70 patients with profound SNHL for possible cochlear implantation surgery followed-up by Helwan University Teaching Hospitals between January 2016 and December 2017.All patients were subjected to high resolution CT scan (64 mutlislice) and 3D T2 MRI of the temporal bone to delineate condition of the mastoid, vascular anomaly, internal auditory canal, cochlear nerve, size of the cochlea, status of the endo-and perilymphatic fluid, round window niche and variation of the facial nerve.All Patients were Classified into: Absolute contraindica-tions to cochlear implant surgery, relative contraindications to cochlear implant surgery and their complicating factor in making surgical intervention with relative difficulty, otherwise easy surgical intervention is assumed.Results: We studied a total of 70 patients with profound deafness planned for cochlear implantation surgery and our results showed 5.7% (n=4) of the study group would not benefit and cochlear implantation surgery is contraindicated for these patients as well as 41.4% (n=29) of the patients have had complicating criteria that possibly make surgical inter-vention difficult in comparison 52.9% (n=37) with anticipated straight forward surgical intervention which were statistically significant.Conclusion: The correct classification of cochlear condi-tions and a clear description of such abnormalities by means of multislice CT and 3D MRI are determining factors in the surgical planning developed by the cochlear implantation team, with direct impact on the success of the surgical inter-vention. Thus the radiologist experienced in the evaluation of the temporal bone plays a major role in the course of this disorder.

A model and experimental approach to the middle ear transfer function related to hearing in the humpback whale (Megaptera novaeangliae).

At present, there are no direct measures of hearing for any baleen whale (Mysticeti). The most viable alternative to in vivo approaches to simulate the audiogram is through modeling outer, middle, and inner ear functions based on the anatomy and material properties of each component. This paper describes a finite element model of the middle ear for the humpback whale (Megaptera novaeangliae) to calculate the middle ear transfer function (METF) to determine acoustic energy transmission to the cochlea. The model was developed based on high resolution computed tomography imaging and direct anatomical measurements of the middle ear components for this mysticete species. Mechanical properties for the middle ear tissues were determined from experimental measurements and published values. The METF for the humpback whale predicted a better frequency range between approximately 15 Hz and 3 kHz or between 200 Hz and 9 kHz based on two potential stimulation locations. Experimental measures of the ossicular chain, tympanic membrane, and tympanic bone velocities showed frequency response characteristics consistent with the model. The predicted best sensitivity hearing ranges match well with known vocalizations of this species.

Identification of emphysema patterns in high resolution computed tomography images

Fractal dimension (FD) is a very useful metric for the analysis of image structures with statistically self-similar properties. It has applications in areas such as texture segmentation, shape classification and analysis of medical images. Several approaches can be used for calculating the fractal dimension of digital images; the most popular method is the box-counting method. It is also very challenging and difficult to classify patterns in high resolution computed tomography images (HRCT) using this important descriptor. This paper applied the Holder exponent computation of the local intensity values for detecting the emphysema patterns in HRCT images. The absolute differences between the normal and the abnormal regions in the images are the key for a successful classification of emphysema patterns using the statistical analysis. The results obtained in this paper demonstrated the effectiveness of the predictive power of the features extracted from the Holder exponent in the analysis and classification of HRCT images. The overall classification accuracy achieved in lung tissue layers is greater than 90%, which is an evidence to prove the effectiveness of the methods investigated in this paper.

Accounting for sub-resolution pores in models of water and solute transport in soils based on computed tomography images: Are we there yet?

In the last decade, X-ray computed tomography (CT) has become widely used to characterize the geometry and topology of the pore space of soils and natural porous media. Regardless of the resolution of CT images, a fundamental problem associated with their use, for example as a starting point in simulation efforts, is that sub-resolution pores are not detected. Over the last few years, a particular type of modeling method, known as “Grey” or “Partial Bounce Back” Lattice-Boltzmann (LB), has been adopted by increasing numbers of researchers to try to account for sub-resolution pores in the modeling of water and solute transport in natural porous media. In this short paper, we assess the extent to which Grey LB methods indeed offer a workable solution to the problem at hand. We conclude that, in spite of significant computational advances, a major experimental hurdle related to the evaluation of the penetrability of sub-resolution pores, is blocking the way ahead. This hurdle will need to be cleared before Grey LB can become a credible option in the microscale modeling of soils and sediments. A necessarily interdisciplinary effort, involving both modelers and experimentalists, is needed to clear the path forward.