Robust Machine Learning Technique to Classify COVID-19 Using Fusion of Texture and Vesselness of X-Ray Images

Manual investigation of chest radiography(CXR)images by physicians is crucial for effective decision-making in COVID-19 diagnosis.However,the high demand during the pandemic necessitates auxiliary help through image analysis and machine learning techniques.This study presents a multi-threshold-based segmentation technique to probe high pixel intensity regions in CXR images of various pathologies,including normal cases.Texture information is extracted using gray co-occurrence matrix(GLCM)-based features,while vessel-like features are obtained using Frangi,Sato,and Meijering filters.Machine learning models employing Decision Tree(DT)and RandomForest(RF)approaches are designed to categorize CXR images into common lung infections,lung opacity(LO),COVID-19,and viral pneumonia(VP).The results demonstrate that the fusion of texture and vesselbased features provides an effective ML model for aiding diagnosis.The ML model validation using performance measures,including an accuracy of approximately 91.8%with an RF-based classifier,supports the usefulness of the feature set and classifier model in categorizing the four different pathologies.Furthermore,the study investigates the importance of the devised features in identifying the underlying pathology and incorporates histogrambased analysis.This analysis reveals varying natural pixel distributions in CXR images belonging to the normal,COVID-19,LO,and VP groups,motivating the incorporation of additional features such as mean,standard deviation,skewness,and percentile based on the filtered images.Notably,the study achieves a considerable improvement in categorizing COVID-19 from LO,with a true positive rate of 97%,further substantiating the effectiveness of the methodology implemented.

A machine learning-based strategy for predicting the mechanical strength of coral reef limestone using X-ray computed tomography

Different sedimentary zones in coral reefs lead to significant anisotropy in the pore structure of coral reef limestone(CRL),making it difficult to study mechanical behaviors.With X-ray computed tomography(CT),112 CRL samples were utilized for training the support vector machine(SVM)-,random forest(RF)-,and back propagation neural network(BPNN)-based models,respectively.Simultaneously,the machine learning model was embedded into genetic algorithm(GA)for parameter optimization to effectively predict uniaxial compressive strength(UCS)of CRL.Results indicate that the BPNN model with five hidden layers presents the best training effect in the data set of CRL.The SVM-based model shows a tendency to overfitting in the training set and poor generalization ability in the testing set.The RF-based model is suitable for training CRL samples with large data.Analysis of Pearson correlation coefficient matrix and the percentage increment method of performance metrics shows that the dry density,pore structure,and porosity of CRL are strongly correlated to UCS.However,the P-wave velocity is almost uncorrelated to the UCS,which is significantly distinct from the law for homogenous geomaterials.In addition,the pore tensor proposed in this paper can effectively reflect the pore structure of coral framework limestone(CFL)and coral boulder limestone(CBL),realizing the quantitative characterization of the heterogeneity and anisotropy of pore.The pore tensor provides a feasible idea to establish the relationship between pore structure and mechanical behavior of CRL.

A Comprehensive Investigation of Machine Learning Feature Extraction and ClassificationMethods for Automated Diagnosis of COVID-19 Based on X-ray Images

The quick spread of the CoronavirusDisease(COVID-19)infection around the world considered a real danger for global health.The biological structure and symptoms of COVID-19 are similar to other viral chest maladies,which makes it challenging and a big issue to improve approaches for efficient identification of COVID-19 disease.In this study,an automatic prediction of COVID-19 identification is proposed to automatically discriminate between healthy and COVID-19 infected subjects in X-ray images using two successful moderns are traditional machine learning methods(e.g.,artificial neural network(ANN),support vector machine(SVM),linear kernel and radial basis function(RBF),k-nearest neighbor(k-NN),Decision Tree(DT),andCN2 rule inducer techniques)and deep learningmodels(e.g.,MobileNets V2,ResNet50,GoogleNet,DarkNet andXception).A largeX-ray dataset has been created and developed,namely the COVID-19 vs.Normal(400 healthy cases,and 400 COVID cases).To the best of our knowledge,it is currently the largest publicly accessible COVID-19 dataset with the largest number of X-ray images of confirmed COVID-19 infection cases.Based on the results obtained from the experiments,it can be concluded that all the models performed well,deep learning models had achieved the optimum accuracy of 98.8%in ResNet50 model.In comparison,in traditional machine learning techniques, the SVM demonstrated the best result for an accuracy of 95% and RBFaccuracy 94% for the prediction of coronavirus disease 2019.

Rapid detection and risk assessment of soil contamination at lead smelting site based on machine learning

A general prediction model for seven heavy metals was established using the heavy metal contents of 207soil samples measured by a portable X-ray fluorescence spectrometer(XRF)and six environmental factors as model correction coefficients.The eXtreme Gradient Boosting(XGBoost)model was used to fit the relationship between the content of heavy metals and environment characteristics to evaluate the soil ecological risk of the smelting site.The results demonstrated that the generalized prediction model developed for Pb,Cd,and As was highly accurate with fitted coefficients(R^(2))values of 0.911,0.950,and 0.835,respectively.Topsoil presented the highest ecological risk,and there existed high potential ecological risk at some positions with different depths due to high mobility of Cd.Generally,the application of machine learning significantly increased the accuracy of pXRF measurements,and identified key environmental factors.The adapted potential ecological risk assessment emphasized the need to focus on Pb,Cd,and As in future site remediation efforts.

Monte Carlo simulation for bremsstrahlung and photoneutron yields in high-energy x-ray radiography

This paper reports on the results of calculations using a Monte Carlo code （MCNP5） to study the properties of photons, electrons and photoneutrons obtained in the converted target and their transportations in x-ray radiography. A comparison between measurements and calculations for bremsstrahlung and photoneutrons is presented. The radiographic rule and the effect of the collimator on the image are studied with the experimental model. The results provide exact parameters for the optimal design of radiographic layout and shielding systems.

Flash X-ray radiography technique to study the high velocity impact of soft projectile on E-glass/epoxy composite material

In the present paper, the high velocity impact of 9 mm soft lead projectile on 10 mm and 30 mm thick Eglass/epoxy composites was studied using a 450 kV Flash X-ray radiography(FXR) system. The basic parameters of FXR imaging, such as effect of ratio of target to film(TF) and source to target(ST) distances and X-ray penetration thickness of the composite material were optimized based on clarity and the actual dimensions of the objects. The optimized parameters were used in the FXR imaging of the ballistic event of 9 mm soft projectile on E-glass/epoxy composite. The real time deformation patterns of both the projectile and composite target during the ballistic impact were captured and studied at different time intervals. The notable failure modes of the 10 mm thick target with time include fibre breakage, bulging on the back side, delamination, recovery of the bulging, reverse bulging and its recovery. However, with increase in thickness of the target to 30 mm the only failure mechanism observed is the breaking of fibres. The ballistic impact event was also numerically simulated using commercially available LS-DYNA software. The numerically simulated deformation patterns of the projectile and target at different time intervals are closely matching with the corresponding radiographic images.

Low Energy Plasma Focus as an Intense X-ray Source for Radiography

Study on X-ray emission from a low energy (1.8 kJ) plasma focus devicepowered by a 9 μF capacitor bank, charged at 20 kV and giving peak discharge current of about 175kA by using a lead-inserted copper-tapered anode is reported. The X-ray yield in different energywindows is measured as a function of hydrogen filling pressure. The maximum yield in 4π-geometry isfound to be (27.3+-1.1) J and corresponding wall plug efficiency for X-ray generation is 1.52+-0.06%. X-ray emission, presumably due to bombarding activity of electrons in current sheath at theanode tip was dominant, which is confirmed by the pinhole images. The feasibility of the device asan intense X-ray source for radiography is demonstrated.

X-Pinches as Broadband Sources of X-Rays for Radiography

Two methods of using the X-pinch as a source of X-ray radiation for radiography of biological objects are presented. X-pinches are found to be a very flexible method for generation of radiation over a wide spectral range and provide a high spatial and temporal resolution.

Application of machine learning in oral and maxillofacial surgery

Oral and maxillofacial anatomy is extremely complex,and medical imaging is critical in the diagnosis and treatment of soft and bone tissue lesions.Hence,there exists accumulating imaging data without being properly utilized over the last decades.As a result,problems are emerging regarding how to integrate and interpret a large amount of medical data and alleviate clinicians’workload.Recently,artificial intelligence has been developing rapidly to analyze complex medical data,and machine learning is one of the specific methods of achieving this goal,which is based on a set of algorithms and previous results.Machine learning has been considered useful in assisting early diagnosis,treatment planning,and prognostic estimation through extracting key features and building mathematical models by computers.Over the past decade,machine learning techniques have been applied to the field of oral and maxillofacial surgery and increasingly achieved expert-level performance.Thus,we hold a positive attitude towards developing machine learning for reducing the number of medical errors,improving the quality of patient care,and optimizing clinical decision-making in oral and maxillofacial surgery.In this review,we explore the clinical application of machine learning in maxillofacial cysts and tumors,maxillofacial defect reconstruction,orthognathic surgery,and dental implant and discuss its current problems and solutions.

Radiography Image Classification Using Deep Convolutional Neural Networks

Research has shown that chest radiography images of patients with different diseases, such as pneumonia, COVID-19, SARS, pneumothorax, etc., all exhibit some form of abnormality. Several deep learning techniques can be used to identify each of these anomalies in the chest x-ray images. Convolutional neural networks (CNNs) have shown great success in the fields of image recognition and image classification since there are numerous large-scale annotated image datasets available. The classification of medical images, particularly radiographic images, remains one of the biggest hurdles in medical diagnosis because of the restricted availability of annotated medical images. However, such difficulty can be solved by utilizing several deep learning strategies, including data augmentation and transfer learning. The aim was to build a model that would detect abnormalities in chest x-ray images with the highest probability. To do that, different models were built with different features. While making a CNN model, one of the main tasks is to tune the model by changing the hyperparameters and layers so that the model gives out good training and testing results. In our case, three different models were built, and finally, the last one gave out the best-predicted results. From that last model, we got 98% training accuracy, 84% validation, and 81% testing accuracy. The reason behind the final model giving out the best evaluation scores is that it was a well-fitted model. There was no overfitting or underfitting issues. Our aim with this project was to make a tool using the CNN model in R language, which will help detect abnormalities in radiography images. The tool will be able to detect diseases such as Pneumonia, Covid-19, Effusions, Infiltration, Pneumothorax, and others. Because of its high accuracy, this research chose to use supervised multi-class classification techniques as well as Convolutional Neural Networks (CNNs) to classify different chest x-ray images. CNNs are extremely efficient and successful at reducing the number of parameters while maintaining the quality of the primary model. CNNs are also trained to recognize the edges of various objects in any batch of images. CNNs automatically discover the relevant aspects in labeled data and learn the distinguishing features for each class by themselves.

In situ study on columnar-equiaxed transition and anaxial columnar dendrite growth of Al-15%Cu alloy by synchrotron radiography

Directional solidification of Al-15% （mass fraction） Cu alloy was investigated by in situ and real time radiography which was performed by Shanghai synchrotron radiation facility （SSRF）. The imaging results reveal that columnar to equiaxed transition （CET） is provoked by external thermal disturbance. The detaching and floating of fragments of dendrite arms are the prelude of the transition when the solute boundary layer in front of the solid-liquid interface is thin. And the dendrite triangular tip is the fracture sensitive zone. When the conditions are suitable, new dendrites can sprout and grow up. This kind of dendrite has no obvious stem and is named anaxial columnar dendrites.

X-ray image distortion correction based on SVR

X-ray image has been widely used in many fields such as medical diagnosis,industrial inspection,and so on.Unfortunately,due to the physical characteristics of X-ray and imaging system,distortion of the projected image will happen,which restrict the application of X-ray image,especially in high accuracy fields.Distortion correction can be performed using algorithms that can be classified as global or local according to the method used,both having specific advantages and disadvantages.In this paper,a new global method based on support vector regression（SVR）machine for distortion correction is proposed.In order to test the presented method,a calibration phantom is specially designed for this purpose.A comparison of the proposed method with the traditional global distortion correction techniques is performed.The experimental results show that the proposed correction method performs better than the traditional global one.

Covid-19 Detection from Chest X-Ray Images Using Advanced Deep Learning Techniques

Like the Covid-19 pandemic,smallpox virus infection broke out in the last century,wherein 500 million deaths were reported along with enormous economic loss.But unlike smallpox,the Covid-19 recorded a low exponential infection rate and mortality rate due to advancement inmedical aid and diagnostics.Data analytics,machine learning,and automation techniques can help in early diagnostics and supporting treatments of many reported patients.This paper proposes a robust and efficient methodology for the early detection of COVID-19 from Chest X-Ray scans utilizing enhanced deep learning techniques.Our study suggests that using the Prediction and Deconvolutional Modules in combination with the SSD architecture can improve the performance of the model trained at this task.We used a publicly open CXR image dataset and implemented the detectionmodelwith task-specific pre-processing and near 80:20 split.This achieved a competitive specificity of 0.9474 and a sensibility/accuracy of 0.9597,which shall help better decision-making for various aspects of identification and treat the infection.

Simulation study on characteristics of information extraction in multiple-image radiography

Simulation experiments were performed to investigate the characteristics of information extraction in multiple-image radiography(MIR) based on geometrical optics approximation. Different Poisson noise levels were added to the simulation, and the results show that Poisson noise deteriorates the extraction results, with the degree of refraction > USAXS > absorption. The effects of Poisson noise are negligible when the detector's photon counts are about 1000 ph/pixel. A wider sampling range allows more accurate extraction results, but a narrower sampling range has a better signal-to-noise ratio for high Poisson noise levels, e.g., PN(10). The sampling interval can be suitably increased with a minor impact on the extraction results for low Poisson noise levels(PN(10000)). The extraction results are incomplete because a portion of the samplerocking curve is beyond the sampling range. This induces artifacts in the images, especially for strong refraction and USAXS signals. The artifacts are not obvious when the refraction angle and standard deviation of the USAXS are smaller than the sampling range by an order of magnitude.In general, the absorption barely affects the extraction results. However, additional Poisson noise will be generated when the sample is made of high-Z elements or has a large size due to the strong absorption. Here, the extraction results will deteriorate, and additional exposure time is required. This simulation provides important details on practical applications of MIR, with improvements in information extraction.

Enhanced K-Edge Radiography Using a High-Spatial-Resolution Cadmium Telluride Array Detector

To confirm the imaging effect of a dual-energy (DE) cadmium telluride (CdTe) array detector (XCounter, Actaeon) and to perform fundamental studies on DE computed tomography, we performed enhanced K-edge radiography using iodine (I) and gadolinium (Gd) media. DE radiography was performed using an X-ray generator with a 0.1-mm-diam-focus tube and a 0.5-mm-thick beryllium window, a 1.0-mm-thick aluminum filter for absorbing extremely low-energy photons, and the CdTe array detector with pixel dimensions of 0.1 × 0.1 mm2. Each pixel has a charge-sensitive amplifier and a dual-energy counter, and the event pulses from the amplifier are sent to the counter to determine two threshold energies. The tube current was a maximum value of 0.50 mA, and the tube voltages for I- and Gd-K-edge radiograms were 60 and 80 kV, respectively. In the I-K-edge radiography of a dog-heart phantom at an energy range of 33 - 60 keV, the muscle density increased, and fine coronary arteries were visible. Utilizing Gd-K-edge radiography of a rabbit head phantom at an energy range of 50 - 80 keV, the muscle density increased, and fine blood vessels in the nose were observed at high contrasts. Using the DE array detector, we confirmed the image-contrast variations with changes in the threshold energy.

A Derived Exposure Chart for Computed Radiography in a Negroid Population

Background: Computed radiography has a wider exposure latitude when compared with film-screen imaging system. Consequently, the risk of dose creep is high. A conscientious effort is there-fore, needed by the radiographer to keep exposure as low as reasonably achievable. Objective: To derive a computed radiography exposure chart for a negroid population using AGFA photostimulable phosphor plates and a GE static X-ray machine. Materials and Method: A static X-ray machine, a digitizer, and photostimulable phosphor plates were used for the X-ray examination. Chest examinations were done at a Focus-Film-Distance (FFD) of 150 - 180 cm while all other examinations were conducted at 90 - 100 cm FFD. The range of exposure factors (kVp, mA and mAs) used by radiog-raphers in the centre was noted and the 90th percentile calculated. Over a three-month period, the patients were examined with the 90th percentile of tube potential (kVp) while keeping other factors constant. The kVp was gradually decreased and halted if radiologists and radiographers uncon-nected with the work expressed misgivings about the quality of the image. A similar procedure was adopted for the tube current (mA). The threshold adopted as low as reasonably achievable was the factor preceding the point of observation by other personnel. Metrics for central tendency from the statistical packages for social sciences, version 17.0 was used to analyze the data. Results: 335 subjects of both gender aged 0 - 92 years were examined by the researchers. Adult exposure factors used by the radiographers (and those derived by the researchers) had a range of 45 - 130 kVp (62 - 94 kVp), 63 - 320 mA (100 - 250 mA) and 4.0 - 25.0 mAs (5.0 - 20.0 mAs) respectively. Pediatric chest (and researchers-derived) factors were 50 - 75 kVp (52 - 65 kVp), 50 - 250 mA (100 - 220 mA) and 3.20 - 10.0 mAs (3.2 - 6.5 mAs) respectively. Conclusion: Upper threshold of adult (and paediatric) exposure factors in computed radiography with comparable equipment and accessories should not exceed 94 kVp (65 kVp), 250 mA (220 mA) and 20.0 mAs (6.5 mAs) respectively. The derived exposure chart is also adequate to address motion unsharpness in chest examinations.

Economic Evaluation of Conventional Radiography with Film and Computed Radiography: Applied at BMC

Conventional radiography with film (CRF) has been in use for diagnostic purposes for a long time now. It has proved to be a great assert for the radiographers in assessing various abnormalities. With recent advances in technology it is now possible to have digital solutions for radiography problems at a very cost effective, environment friendly and also with better image quality in certain applications when compared to CRF. Rather than using a CRF a computed radiography (CR) uses imaging plates to capture the image. The imaging plate contains photosensitive phosphors which contain the latent image. Later this plate is introduced into a reader which is then converted into a digital image. The major advantage and the cost effective element of this system is the ability to reuse the imaging plates unlike the photographic film where in only a single image can be captured and cannot be reused. The computed radiography drastically reduces the cost by eliminating the use of chemicals like film developers and fixers and also the need for a storage room. It also helps to reduce the costs that are involved in the disposal of wastes due to conventional radiography. This paper investigates whether it is cost effective to use computed radiography over film based system at Al-Batnan Medical Center (BMC), Tobruk, Libya by using Cost Benefit Analysis (CBA). Apart from the initial cost of the CR System, based on the data collected from the center, from the year 2008 to 2012 (until June 2012) a total of 581,566 images were produced with the total cost incurred using film based system being USD 4,652,528. If the same number of images were produced using a CR system the total cost incurred would have been USD 82,600. Taking into consideration the cost of a new CR system to be USD 120,000 the overall cost of producing these images is USD 202,600. It is observed that an amount of USD 4,449,928 could have been saved over the period of 5 years starting from 2008 to 2012 by using the CR system at BMC. Using Cost Benefit Analysis, the average value of the net difference between the costs and benefits for the conventional film based system is ?83.38 where as for the Computed System it is 22.06. Based on the principles of Cost Benefit Analysis it can be concluded that the system with a net positive difference is more cost beneficial than the other. With the help of the above two analysis it can be concluded that the use of computed radiography is definitely more cost effective for use at BMC, when compared to the conventional x-ray radiography.

Multi-Label Chest X-Ray Classification via Deep Learning

In this era of pandemic, the future of healthcare industry has never been more exciting. Artificial intelligence and machine learning (AI & ML) present opportunities to develop solutions that cater for very specific needs within the industry. Deep learning in healthcare had become incredibly powerful for supporting clinics and in transforming patient care in general. Deep learning is increasingly being applied for the detection of clinically important features in the images beyond what can be perceived by the naked human eye. Chest X-ray images are one of the most common clinical method for diagnosing a number of diseases such as pneumonia, lung cancer and many other abnormalities like lesions and fractures. Proper diagnosis of a disease from X-ray images is often challenging task for even expert radiologists and there is a growing need for computerized support systems due to the large amount of information encoded in X-Ray images. The goal of this paper is to develop a lightweight solution to detect 14 different chest conditions from an X ray image. Given an X-ray image as input, our classifier outputs a label vector indicating which of 14 disease classes does the image fall into. Along with the image features, we are also going to use non-image features available in the data such as X-ray view type, age, gender etc. The original study conducted Stanford ML Group is our base line. Original study focuses on predicting 5 diseases. Our aim is to improve upon previous work, expand prediction to 14 diseases and provide insight for future chest radiography research.

Development of a monochromatic crystal backlight imager for the recent double-cone ignition experiments

We developed a monochromatic crystal backlight imaging system for the double-cone ignition(DCI) scheme, employing a spherically bent quartz crystal. This system was used to measure the spatial distribution and temporal evolution of the head-on colliding plasma from the two compressing cones in the DCI experiments. The influence of laser parameters on the x-ray backlighter intensity and spatial resolution of the imaging system was investigated. The imaging system had a spatial resolution of 10 μm when employing a CCD detector. Experiments demonstrated that the system can obtain time-resolved radiographic images with high quality, enabling the precise measurement of the shape, size, and density distribution of the plasma.

Characterization of the Convoluted 3D Internetallic Phases in a Recycled Al Alloy by Synchrotron X-ray Tomography and Machine Learning

Fe-rich intermetallic phases in recycled Al alloys often exhibit complex and 3D convoluted structures and morphologies.They are the common detrimental intermetallic phases to the mechanical properties of recycled Al alloys.In this study,we used synchrotron X-ray tomography to study the true 3D morphologies of the Ferich phases,Al_(2)Cu phases and casting defects in an ascast Al-5Cu-1.5Fe-1Si alloy.Machine learning-based image processing approach was used to recognize and segment the diff erent phases in the 3D tomography image stacks.In the studied condition,theβ-Al_(9)Fe_(2)Si_(2)andω-Al_(7)Cu_(2)Fe are found to be the main Fe-rich intermetallic phases.Theβ-Al_(9)Fe_(2)Si_(2)phases exhibit a spatially connected 3D network structure and morphology which in turn control the 3D spatial distribution of the Al_(2)Cu phases and the shrinkage cavities.The Al_(3)Fe phases formed at the early stage of solidification aff ect to a large extent the structure and morphology of the subsequently formed Fe-rich intermetallic phases.The machine learning method has been demonstrated as a powerful tool for processing big datasets in multidimensional imaging-based materials characterization work.