Multicontext 3D residual CNN for false positive reduction of pulmonary nodule detection

False positive reduction is a key procedure of computer-aided pulmonary nodule detection. The goal is to recognize the true pulmonary nodule from the plentiful candidates, which received from the first step of pulmonary nodule candidate detection. Convolutional networks can be used to perform false positive nodule reduction, but the classification accuracy need to be further improved. Recently, residual network is more and more popular around the world with its distinguished performance. A multicontext three-dimensional residual convolutional neural network (3D Res-CNN) was proposed to realize the reduction of the false positive nodule. Using two scales of network to adapt the variation of pulmonary nodule size, instead of using an unreferenced function with reference to the identity mapping, 3D Res-CNN uses a shortcut connection to realize the residual structure. For alleviating the data imbalance, firstly patches are rotated and resampled in original images; secondly weights are allotted for different labels in the calculation of cost function. Experiments on volumetric computed tomography (CT) data indicate that our method gets state of the art performance: 0.843 average sensitivity with 0.125, 0.25, 0.5, 1, 2, 4, and 8 false positive per subject. The results show the effectiveness of residual convolutional network for the recognition of the true pulmonary nodule from the plentiful candidates.     

Dynamic modulus simulation of the asphalt concrete using the X-ray computed tomography images

The objective of this study is to predict the dynamic modulus of asphalt mixture using both two-dimensional (2D) and three-dimensional (3D) Distinct Element Method (DEM) generated from the X-ray computed tomography (X-ray CT) images. The 3D internal microstructure of the asphalt mixtures (i.e., spatial distribution of aggregate, sand mastic and air voids) was obtained using the X-ray CT. The X-ray CT images provided exact locations of aggregate, sand mastic and air voids to develop 2D and 3D models. An experimental program was developed with a uniaxial compression test to measure the dynamic modulus of sand mastic and asphalt mixtures at different temperatures and loading frequencies. In the DEM simulation, the mastic dynamic modulus and aggregate elastic modulus were used as input parameters to predict the asphalt mixture dynamic modulus. Three replicates of a 3D DEM and six replicates of a 2D DEM were used in the simulation. The strain response of the asphalt concrete under a compressive load was monitored, and the dynamic modulus was computed. The moduli of the 3D DEM and 2D DEM were then compared with both the experimental measurements results. It was revealed that the 3D discrete element models successfully predicted the asphalt mixture dynamic modulus over a range of temperatures and loading frequencies. It was found that 2D discrete element models under predicted the asphalt mixture dynamic modulus.     

High-resolution 3D imaging of microvascular architecture in human glioma tissues using X-ray phase-contrast computed tomography as a potential adjunct to histopathology

Gliomas are rich in blood vessels, and the generation of tumor-associated vessels plays an important role in glioma growth and transfer. Histology can directly depict microvascular architecture in the tumor, but it just provides two-dimensional (2D) images obtained by destroying three-dimensional (3D) tissue specimens. There is a lack of high-resolution 3D imaging methods for observing the microvasculature throughout the entire specimens. X-ray phase-contrast computed tomography (PCCT) which is an emerging imaging method has demonstrated its outstanding potential in imaging soft tissues. Thus, this study aims to evaluate the potential of PCCT as an adjunct to histopathology in nondestructive and 3D visualization of the microvascular architecture in human glioma tissues. In this study, seven resected glioma tissues were scanned via PCCT and then processed histologically. The obtained PCCT data was analyzed and compared with corresponding histological results. Significant anatomical structures of the glioma such as microvessels, thrombi inside the microvessels, and areas of vascular proliferation could be clearly presented via PCCT, confirmed by the histological findings. Moreover, PCCT data also provided additional 3D information such as morphological alterations of the microvasculature, 3D distribution of the thrombi and stenosis severity of the vessels in glioma tissues, which cannot be fully analyzed in 2D histological slices. In conclusion, this study demonstrated that PCCT can offer excellent images at a near-histological level and additional valuable information in screening gliomas, without impeding further histological investigations. Thus, this technique could be potentially used as an adjunct to conventional histopathology in 3D nondestructive characterization of glioma vasculature.     

Thyroid nodules classification and diagnosis in ultrasound images using fine-tuning deep convolutional neural network

Ultrasonography AKA diagnostic sonography is a noninvasive imaging technique that allows the analysis of an organic structure, thanks to the ultrasonic waves. It is a valuable diagnosis method and is also seen as the evidence-based diagnostic method for thyroid nodules. The diagnosis, however, is visually made by the practitioner. The automatic discrimination of benign and malignant nodules would be very useful to report Thyroid Imaging Reporting. In this paper, we propose a fine-tuning approach based on deep learning using a Convolutional Neural Network model named resNet-50. This approach allows improving the effectiveness of the classification of thyroid nodules in ultrasound images. Experiments have been conducted on 814 ultrasound images and the results show that our proposed approach dramatically improves the accuracy of the classification of thyroid nodules and outperforms The VGG-19 model.     

In-situ X-ray computed tomography characterisation of 3D fracture evolution and image-based numerical homogenisation of concrete

In-situ micro X-ray Computed Tomography (XCT) tests of concrete cubes under progressive compressive loading were carried out to study 3D fracture evolution. Both direct segmentation of the tomography and digital volume correlation (DVC) mapping of the displacement field were used to characterise the fracture evolution. Realistic XCT-image based finite element (FE) models under periodic boundaries were built for asymptotic homogenisation of elastic properties of the concrete cube with Young's moduli of cement and aggregates measured by micro-indentation tests. It is found that the elastic moduli obtained from the DVC analysis and the FE homogenisation are comparable and both within the Reuss-Voigt theoretical bounds, and these advanced techniques (in-situ XCT, DVC, micro-indentation and image-based simulations) offer highly-accurate, complementary functionalities for both qualitative understanding of complex 3D damage and fracture evolution and quantitative evaluation of key material properties of concrete.     

Automatic liver tumor segmentation on multiphase computed tomography volume using SegNet deep neural network and K-means clustering

Liver and liver tumor segmentations are essential in computer-aided systems for diagnosing liver tumors. These systems must operate on multiphase computed tomography (CT) images instead of a single phase for accurate diagnosis for clinical applications. We have proposed a framework that can perform segmentation from quadriphasic CT data. The liver was segmented using a fine-tuned SegNet model and the liver tumor was segmented using the K-means clustering method coupled with a power-law transformation-based image enhancement technique. The best values for liver segmentation achieved were: Dice Coefficient = 96.46 ± 0.48%, Jaccard Index = 93.16 ± 0.89%, volumetric overlap error = 6.84 ± 0.89% and average symmetric surface distance = 0.59 ± 0.3 mm and the results for liver tumor delineation were Dice Coefficient = 85.07 ± 4.5%, Jaccard Index = 74.29 ± 6.8%, volumetric overlap error = 25.71 ± 6.8% and average symmetric surface distance = 1.14 ± 1.3 mm. The proposed liver segmentation method based on deep learning is fully automatic, robust, and effective for all phases. The image enhancement technique has shown promising results and aided in better liver tumor segmentation. The liver tumors were segmented satisfactorily; however, improvements concerning false positive reduction can further increase the accuracy.     

Geometrical approach for automatic detection of liquid surfaces in 3D computed tomography baggage imagery

This study presents a novel method for liquid detection within three-dimensional (3D) computed tomography (CT) baggage inspection imagery. Liquid detection within airport security is currently of significant interest due to security threats associated with liquid explosives. In this paper, we propose a robust technique based on the automatic identification of universal geometric properties of liquids within 3D space. The proposed approach is based on two stages of geometric fitting. First, we identify the 3D plane which fits to the horizontally oriented surface of the liquid recognising the universal self-levelling property of liquids in any given container. Second, we conduct two-dimensional shape analysis to highlight the shape of the liquid surface at a given level within the container using a least squares elliptical fitting approach. The proposed approach relies on the fact that occurrences of such perfectly aligned horizontal planes within a 3D CT security baggage scan are generally unlikely. Occurrences of such instance are thus indicative of liquid presence. Our results, over an extended set of complex test examples, confirm a liquid detection rate of 85–98% with a moderate processing time. Furthermore, as this proposed approach is based purely on the geometric properties of liquids and robust geometrical shape detection, this methodology is intrinsic to the 3D nature of the resulting CT data and not dependent on any exemplar training imagery.     

Non-invasive and rapid analysis for observation of internal structure of press-coated tablet using X-ray computed tomography

Background: Since the internal structure of a tablet can be measured without destruction of the sample by X‐ray computed tomography (CT), it could be applied to quality control of tablets during the manufacturing process. Aim: A novel, fast, noninvasive tablet observation method was developed to evaluate the internal structure of commercial press-coated tablets by using X-ray CT. Method: Thirty-two CT image slices of four kinds of commercial press-coated tablets (tablets A, B, C, and D) were measured 300 m interval between edges of the tablet by using an X-ray CT. The thinnest layer thickness of the tablets and distance between centers of gravity (DCG) of tables were calculated. Results: The order of the TLT of the tablets was tablet B > tablet C > tablet D > tablet A. The result indicated that the order of DCG was tablet A > tablet D > tablet C > tablet B. Noninvasive observation of the internal structure of commercial, press-coated tablets by X-ray CT has been demonstrated to be useful in quality control of production. Conclusion: The internal structure of press-coated tablets could be observed without pretreatment, without destruction, and very rapidly by X‐ray CT.     

Automated segmentation of synchrotron radiation micro-computed tomography biomedical images using Graph Cuts and neural networks

Synchrotron Radiation (SR) X-ray micro-Computed Tomography (μCT) enables magnified images to be used as a non-invasive and non-destructive technique with a high space resolution for the qualitative and quantitative analyses of biomedical samples. The research on applications of segmentation algorithms to SR-μCT is an open problem, due to the interesting and well-known characteristics of SR images for visualization, such as the high resolution and the phase contrast effect. In this article, we describe and assess the application of the Energy Minimization via Graph Cuts (EMvGC) algorithm for the segmentation of SR-μCT biomedical images acquired at the Synchrotron Radiation for MEdical Physics (SYRMEP) beam line at the Elettra Laboratory (Trieste, Italy). We also propose a method using EMvGC with Artificial Neural Networks (EMANNs) for correcting misclassifications due to intensity variation of phase contrast, which are important effects and sometimes indispensable in certain biomedical applications, although they impair the segmentation provided by conventional techniques. Results demonstrate considerable success in the segmentation of SR-μCT biomedical images, with average Dice Similarity Coefficient 99.88% for bony tissue in Wistar Rats rib samples (EMvGC), as well as 98.95% and 98.02% for scans of Rhodnius prolixus insect samples (Chagas's disease vector) with EMANNs, in relation to manual segmentation. The techniques EMvGC and EMANNs cope with the task of performing segmentation in images with the intensity variation due to phase contrast effects, presenting a superior performance in comparison to conventional segmentation techniques based on thresholding and linear/nonlinear image filtering, which is also discussed in the present article.     

Deep chest X‐ray: Detection and classification of lesions based on deep convolutional neural networks

We investigated whether a convolutional neural network (CNN) can enhance the usability of computer‐aided detection (CAD) of chest radiographs for various pulmonary abnormal lesions. The numbers of normal and abnormal patients were 6055 and 3463, respectively. Two radiologists delineated regions of interest for lesions and labeled the disease types as ground truths. The datasets were split into training, tuning, and testing as 7:1: 2. Total test sets were randomly selected in 1214 normal and 690 abnormal. A 5‐fold, cross‐validation was performed on our datasets. For the classification of normal and abnormal, we developed a CNN based on DenseNet169; for abnormal detection, The You Only Look Once (YOLO) v2 with DenseNet was used. Detection and classification of normal and five classes of diseases (nodule[s], consolidation, interstitial opacity, pleural effusion, and pneumothorax) on chest radiographs were analyzed. Our CNN model classified chest radiographs as normal or abnormal with an accuracy of 97.8%. For the results of the abnormal, F1 score, was 75.2 ± 2.28% for nodules, 55.0 ± 4.3% for consolidation, 78.2 ± 7.85% for interstitial opacity, 81.6 ± 2.07% for pleural effusion, and 70.0 ± 7.97% for pneumothorax, respectively. In addition, we conducted the experiments between our method and RetinaNet with only nodules. The results of our method and RetinaNet at cutoff‐0.5 in the free response operating characteristic curve were 83.45% and 80.55%, respectively. Our algorithm demonstrated viable detection and disease classification capacity and could be used for CAD of lung diseases on chest radiographs.     

Development of a three-dimensional finite element model for a unidirectional carbon fiber reinforced plastic based on X-ray computed tomography images and the numerical simulation on compression

A unidirectional carbon fiber reinforced plastic (CFRP) was scanned by an X-ray computed tomography (CT) system. Based on the X-ray CT images, a three-dimensional model with random fiber waviness was developed. Each fiber location was identified in a sectional CT image. Subsequently, the relative displacement of fibers between adjacent sectional CT images was obtained with a digital image correlation method. This procedure was repeated to obtain fiber waviness along the axial direction. The constructed three-dimensional fiber model showed random waviness of each fiber in the unidirectional CFRP. Finite element analysis was performed using the three-dimensional model. Simulation results showed bending and twisting deformations coupled with axial contractions during axial compression, which developed due to fiber waviness. A reduction of the fiber directional Young’s modulus due to fiber random waviness was quantitatively evaluated.     

Freeze-thaw crack determination in cementitious materials using 3D X-ray computed tomography and acoustic emission

As concrete freezes and thaws cracks may develop. These cracks can provide a path for water and ionic species to penetrate the concrete. This may reduce the service-life of the concrete element. In this study, X-ray computed tomography (CT) was used as a non-destructive technique to characterize the microstructure of mortar samples that were exposed to different levels of freeze-thaw damage by varying degree of saturation in the samples (75, 90, 95, and 100% degrees of saturation). Acoustic emission (AE) experiments were performed during freezing and thawing to investigate sample cracking behavior. The volume of cracks present within the mortar samples after freezing and thawing were determined using X-ray CT and compared to passive acoustic emission data. The location/source of cracks was also determined using X-ray CT. The crack sources (i.e., void, aggregate, interfacial transition zone, or paste) were determined using X-ray CT and were related to AE activities during cracking. Crack volumes were found to increase with increased levels of saturation, and visual observations of cracking were found to correlate with AE signatures of various crack sources.     

Three-dimensional assessment of low velocity impact damage in particle toughened composite laminates using micro-focus X-ray computed tomography and synchrotron radiation laminography

Results are presented studying the contribution of particle toughening to impact damage resistance in carbon fibre reinforced polymer materials. Micro-focus X-ray computed tomography and synchrotron radiation computed laminography were used to provide a novel, multiscale approach for assessing impact damage. Thin (1 mm thick) composite plates containing either untoughened or particle-toughened resin systems were subjected to low velocity impact. Damage was assessed three-dimensionally at voxel resolutions of 0.7 μm and 4.3 μm using SRCL and μCT respectively; the former being an innovative approach to the laterally extended geometry of CFRP plates. Observations and measurements taken from μCT scans captured the full extent of impact damage on both material systems revealing an interconnected network of intra- and inter-laminar cracks. These lower resolution images reveal that the particle-toughened system suppresses delaminations with little effect on intralaminar damage. The higher resolution images reveal that the particles contribute to toughening by crack deflection and bridging.     

3D analytical mathematical models of random star-shape particles via a combination of X-ray computed microtomography and spherical harmonic analysis

To compute any physical quantity for a random particle, one needs to know the mathematical shape of the particle. For regular particles like spheres and ellipsoids, the mathematics are straightforward. For random particles, with realistic shapes, mathematically characterizing the shape had not been generally done. But since about the year 2002, a method has been developed that combines X-ray computed tomography and spherical harmonic analysis to give analytical, differentiable mathematical functions for the three-dimensional shape of star-shape particles, which are a wide class of particles covering most industrial particles of interest, ranging from micrometer scale to millimeter scale particles. This review article describes how this is done, in some detail, and then gives examples of applications of this method, including a contact function that is suitable for these random shape particles. The purpose of this article is to make these ideas widely available for the general powder researcher who knows that particle shape is important to his/her applications, and especially for those researchers who are just starting out in their particle science and technology careers.     

Image fusion methodology for efficient interpretation of multiband images in 3D high‐resolution ultrasonic transmission tomography

With the critical innovations of using submillimeter transducers and multiband analysis of the first arrival pulse, a high‐resolution ultrasonic transmission tomography (HUTT) system has been built and tested to produce multiband images of biological organs at submillimeter resolution. Since the resulting multiband images consist of frequency‐dependent attenuation coefficients (relative to water reference) of transmitted ultrasound pulses, their contrast and sharpness depend on the specific frequency band(s) used for image formation. Even though this multiband representation provides a powerful tool for soft‐tissue differentiation, it hinders visual inspection and limits the visual interpretation of image contents in a short time. To facilitate the visual interpretation of HUTT multiband images, this article presents an efficient image fusion methodology called local principal component analysis with structure tensor (LPCA‐ST). The LPCA has been known as a feasible tool for the fusion of spectral data, since it utilizes the principal components of spectral data as a fusion‐weighting vector of local area. Nonetheless, the LPCA‐fused image often suffers from oversmoothness because of the redundancy of the spectral data. To prevent this problem, we propose a structure tensor as the metric used to select the most informative bands for subsequent LPCA fusion. Our preliminary studies have shown that the contrast of the LPCA‐fused image improves dramatically only when multiband images whose values of the respective structure tensor are the highest are used in the LPCA fusion process. This is achieved in 3D without increasing the computational complexity of the fusion process. © 2009 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 19, 277–282, 2009     

Analysis of cerebral blood flow in Parkinson's disease with dementia versus subcortical ischemic vascular dementia using single photon emission computed tomography

Since both Parkinson's disease with dementia (PDD) and subcortical ischemic vascular dementia (SIVaD) are subcortical dementia syndromes and have similar patterns of cognitive dysfunction, it is difficult to accurately differentiate between these in their early phases using neuropsychological tests. The purpose of this study was to investigate differences in the cerebral perfusion pattern of patients with SIVaD and PDD at the earliest stages using single photon emission computed tomography (SPECT). We, consecutively, recruited 24 patients with mild PDD, 28 patients with mild probable AD, and 33 age‐matched healthy subjects. All subjects underwent Tc‐99m HMPAO perfusion SPECT and general neuropsychological tests. Brain SPECT images were analyzed using the statistical parametric mapping program. There was more significant hypoperfusion in the right cuneus, fusiform gyrus and lingual gyrus in the occipital lobe, right middle temporal gyrus, right postcentral gyrus, and right cerebellar tonsil in PDD compared with the SIVaD group. Conversely, significant hypoperfusion was observed in the bilateral brain stem, limbic system and posterior cingulate gyrus in SIVaD compared with the PDD group. This study suggests that parieto‐occipital hypoperfusion in PDD and hypoperfusion of the brain stem and limbic system in SIVaD are likely useful to differentiate between mild PDD and SIVaD.     

Extraction of stress intensity factors for 3D small fatigue cracks using digital volume correlation and X-ray tomography

This paper describes a methodology used to compute stress intensity factor values along the curved front of a fatigue crack inside a nodular cast iron. An artificial defect is introduced at the surface of a small sample. The initiation and growth of a fatigue crack from this defect during constant amplitude cycling is monitored in situ by laboratory X-ray tomography. The method for processing the 3D images in order to compute SIF values is described in detail. The results obtained show variations of the stress intensity factor values along the crack front.     

SXR-CT技术应用研究烧结陶瓷三维微结构拓扑形貌

借助同步辐射硬X射线高强度，强穿透，高分辨和天然准直性的特性，应用卷积反投影重建算法实现CT三维重建技术，简称SXR—CT(Synchrotron X-Ray Computed Tomography),研究了氧化物陶瓷烧结体残余孔隙的三维拓扑微结构。同时量化计算了样品的孔隙率和密度，较好地符合实验测量数据。     

Detection of bearing defects in three-phase induction motors using Park’s transform and radial basis function neural networks

This paper investigates the application of induction motor stator current signature analysis (MCSA) using Park’s transform for the detection of rolling element bearing damages in three-phase induction motor. The paper first discusses bearing faults and Park’s transform, and then gives a brief overview of the radial basis function (RBF) neural networks algorithm. Finally, system information and the experimental results are presented. Data acquisition and Park’s transform algorithm are achieved by using LabVIEW and the neural network algorithm is achieved by using MATLAB programming language. Experimental results show that it is possible to detect bearing damage in induction motors using an ANN algorithm.