Medical X-ray Imaging Research Articles

Learning from imbalanced COVID-19 chest X-ray (CXR) medical imaging data.

The trendy task of digital medical image analysis has been continually evolving. It has been an area of prominent and growing importance from both research and deployment perspectives. Nonetheless, it is necessary to realize that the use of algorithms, methodology, as well as the source of medical image data, must be strictly scrutinized. As the COVID-19 pandemic has been gripping much of the world recently, there has been much efforts gone into developing affordable testing for the masses, and it has been shown that the established and widely available chest X-rays (CXR) images may be used as a screening criteria for assistive diagnosis purpose. Thanks to the dedicated work by various individuals and organizations, publicly available CXR of COVID-19 subjects are available for analytic usage. We have also provided a publicly available CXR dataset on the Kaggle platform. As a case study, this paper presents a systematic approach to learn from a typically imbalanced set of CXR images, which consists of a limited number of publicly available COVID-19 images. Our results show that we are able to outperform the top finishers in a related Kaggle multi-class CXR challenge. The proposed methodology should be able to help guide medical personnel in obtaining a robust diagnosis model to discern COVID-19 from other conditions confidently.

Extension of the R–L algorithm for [formula omitted]-ray spectra deconvolution and its experimental results

The Richardson–Lucy (R–L) algorithm is widely used for γ-ray spectra deconvolution. In this study, we considered the R–L algorithm as an iteration process of an optimization problem and extended it with a universal function. Three iterative formulae are given based on common elementary functions: the exponential R–L algorithm, the logarithmic R–L algorithm, and the power R–L algorithm. Simulated and experimental 152Eu spectra were used to test the effects of the proposed algorithms. Results show that all the three algorithms make the objective function value decrease monotonically with the increase of iteration number. With proper parameters, the extended algorithms achieve faster convergence than the original R–L algorithm and the MAP algorithm. The power R–L algorithm with m less than but close to 2 shows the fastest convergence rate. This indicates the proposed extension is an effective improvement. Also, the method could be used for other objective functions, which provides a new idea for constructing better iterative algorithms for spectrum or image deconvolution, such as chromatography, infrared spectroscopy, Raman spectroscopy, X-ray fluorescence spectroscopy and medical imaging.

Blind Image Deblurring Using Laplacian of Gaussian (LoG) Based Image Prior

Blind image deconvolution, a technique for obtaining restored image as well as the blur kernel from an inexact image. This research uses spatial characteristics to tackle the problem of blind image deconvolution. To work, the proposed method does not necessitate prior information about the blur kernel. Many applications, such as remote sensing, astronomy, and medical X-ray imaging, necessitate blind image deconvolution algorithms. This study used the maximum a posteriori (MAP) paradigm to create a new blind deblurring approach for removing blur from images. In beginning, we employed a Laplacian of Gaussian (LoG)-based image before regularising the gradients of an image. In the second phase, we used an operator known as the Iterative Shrinkage Thresholding Algorithm (ISTA) to cope with the non-convex challenge that develops during the entire deblurring procedure. Finally, we compared our method to several well-known methods in terms of quantitative and qualitative qualities, and we were able to determine which strategy was the most effective. Our findings show that the strategy we propose outperforms the others by a large margin.

A comparative study of internationally used constancy testing protocols for medical X-ray imaging devices

A comparative study of internationally used constancy testing protocols for medical X-ray imaging devices

On the suitability of XR-SP2 gafchromic films for dosimetry in mammography

XR-SP2 GafChromic films are extensively used in industrial applications. Even if their sensitivity range overlaps with exposure ranges adopted in x-ray medical imaging, there are only few literature examples of their use in medical physics experiments. This work aims at evaluating the XR-SP2 GafChromic film suitability for dosimetric evaluations on 2D mammography systems as well as at providing a complete picture about this film model. The dose-response curve of the film was evaluated using a Hologic Selenia Dimension scanner for W/Rh spectra at 26 kV (air kerma range: 1.40–10.29 mGy) and at 31 kV (air kerma range: 2.15–15.85 mGy). The suitability of linear and non-linear model was evaluated also in comparison to XR-QA2 GafChromic films, adopted as reference. The uncertainty of air kerma measurements with XR-SP2 films decreased as the air kerma increased, down to 4% at 26 kV and down to 5% at 31 kV, evaluated at the upper boundary of the explored exposure range and for non-linear curve fit. In the case of XR-QA2 films, this value tended to 4% both at 26 kV and at 31 kV. Linear model resulted adequate to describe the dose-response curves of the two GafChromic film types, in particular for reduced exposure ranges (air kerma range: 2.60–10.29 mGy @26 kV; 3.98–15.85 mGy @31 kV); the uncertainty analysis did not outline any difference between linear and non-linear curve fit model. From visual assessment, XR-SP2 film lot was observed to present dark dots on the surface that could affect kerma distribution assessments.

Efficient deep learning based data augmentation techniques for enhanced learning on inadequate medical imaging data

<p class="Abstract">The world has come to a standstill with the Coronavirus taking over. In these dire times, there are fewer doctors and more patients and hence, treatment is becoming more and more difficult and expensive. In recent times, Computer Science, Machine Intelligence, measurement technology has made a lot of progress in the field of Medical Science hence aiding the automation of a lot of medical activities. One area of progress in this regard is the automation of the process of detection of respiratory diseases (such as COVID-19). There have been many Convolutional Neural Network (CNN) architectures and approaches that have been proposed for Chest X-Ray Classification. But a big problem still remains and that is the minimal availability of Medical X-Ray Images due to improper measurements. Due to this minimal availability of Chest X-Ray data, most CNN classifiers do not get trained to an optimal level and the required standards for automating the process are not met. In order to overcome this problem, we propose a new deep learning based approach for accurate measurements of physiological data.</p>

Seed-Crystal-Induced Cold Sintering Toward Metal Halide Transparent Ceramic Scintillators.

Scintillators with high spatial resolution at a low radiation dose rate are desirable for X-ray medical imaging. To challenge the state-of-art technology, it is necessary to design large-area wafers with high light yield, oriented light transport, and reduced light scattering. Here, a seed-crystal-induced cold sintering is adopted and a <001>-textured TPP2 MnBr4 (TPP: tetraphenylphosphonium) transparent ceramic is fabricated with a large-area wafer of 5cm in diameter, exhibiting high optical transparency of above 68% over the 450-600nm range. The compelling scintillation performance of the TPP2 MnBr4 wafer includes a light yield of ≈78000± 2000 photons per MeV, a low detection limit 8.8 nanograys per second, about 625 times lower than the requirement of X-ray diagnostics (5500 nanograys per second), and an energy resolution of 17% for high-energy γ-rays (662keV). X-ray imaging demonstrates a high spatial resolution of 15.7 lp mm-1 . Moreover, the designed material exhibits good retention of the radioluminescence intensity and light yield. This work presents a paradigm for achieving light-guiding properties with high transparency and large-area fabrication by grain orientation engineering, and the transparent, textured metal halide ceramic scintillator is expected to provide a route for advancement in the X-ray imaging of tomorrow.

Optimal trajectory based start-up control of LCC resonant converter for X-ray generator applications

LCC series–parallel resonant converter (LCC-SPRC) is the preferred high voltage DC power supply (HVPS) topology for medical X-ray imaging generator applications, due to its advantage in wide operating ranges, and full-range soft switching capability. In this paper, an Input-Single-Output-Series (ISOS) type LCC-SPRC which utilizes the cascaded twelve-level rectifier as the interface with a 140 kV X-ray tube, is presented. However, the existing control strategies cannot meet the requirements of high-voltage generator output voltage response time for real industrial practice. By modeling LCC-SPRC with state plane trajectory analysis (SPTA), the start-up trajectory curve is precisely predicted. By adjusting the current limit band further, the operating mode can be improved to stabilize the initial state, regulating the state variables onto the optimized trajectory locus within the shortest time cycle. Experiment results based on a 140kV/42kW setup are presented to indicate the correctness and feasibility of the proposed method.

X-ray imaging detector for radiological applications adapted to the context and requirements of low- and middle-income countries.

This paper describes the development of a novel medical x-ray imaging system adapted to the needs and constraints of low- and middle-income countries. The developed system is based on an indirect conversion chain: a scintillator plate produces visible light when excited by the x rays, and then, a calibrated multi-camera architecture converts the visible light from the scintillator into a set of digital images. The partial images are then unwarped, enhanced, and stitched through parallel field programmable gate array processing units and specialized software. All the detector components were carefully selected focusing on optimizing the system's image quality, robustness, cost-effectiveness, and capability to work in harsh tropical environments. With this aim, different customized and commercial components were characterized. The resulting detector can generate high quality medical diagnostic images with detective quantum efficiency levels up to 60% (@2.34 μGy), even under harsh environments, i.e., 60 °C and 98% humidity.

Ultralow Detection Limit and Robust Hard X-ray Imaging Detector Based on Inch-Sized Lead-Free Perovskite Cs3Bi2Br9 Single Crystals.

Halide perovskites are promising candidates for soft X-ray detection (<80 keV) owing to their high X-ray absorption coefficient, resistivity, and mobility lifetime product. However, the lack of large high-quality single crystals (SCs) renders it challenging to manufacture robust hard X-ray imaging systems (>100 keV) with a low detection limit and stable dark current. Herein, high-quality inch-size two-dimensional (2D) Cs3Bi2Br9 (CBB) single crystals are grown from a melt via the Bridgman method. The crystal quality is enhanced by eliminating inclusions of CsBr-rich phases and restraining the trap-state density, leading to an enhanced resistivity of 1.41 × 1012 Ω cm and a mobility lifetime product of 8.32 × 10-4 cm2 V-1. The Au/CBB/Au single-crystal device exhibits a high sensitivity of 1705 μC Gyair-1 cm-2 in all-inorganic bismuth-based perovskites and an ultralow detection limit of 0.58 nGyair s-1 in all of the bismuth-based perovskites for 120 keV hard X-ray detection. The CBB detector exhibits high work stability with an ultralow dark current drift of 2.8 × 10-10 nA cm-1 s-1 V-1 and long-term air environment reliability under a high electric field of 10 000 V cm-1 owing to the ultrahigh ionic activation energy of the 2D structure. The proposed robust imaging system based on CBB SC is a promising tool for X-ray medical imaging and diagnostics.

Systematic literature review on the benefit of patient protection shielding during medical X-ray imaging: Towards a discontinuation of the current practice

<h2>Abstract</h2><h3>Purpose</h3> Patient shielding during medical X-ray imaging has been increasingly criticized in the last years due to growing evidence that it often provides minimal benefit and may even compromise image quality. In Europe, and as also shown in a short assessment in Switzerland, the use of patient shielding is inhomogeneous. The aim of this study was to systematically review recent literature in order to assess benefits and appraise disadvantages related to the routine use of patient shielding. <h3>Methods</h3> To evaluate benefits and disadvantages related to the application of patient shielding in radiological procedures, a systematic literature review was performed for CT, radiography, mammography and fluoroscopy-guided medical X-ray imaging. In addition, reports from medical physics societies and authorities of different countries were considered in the evaluation. <h3>Results</h3> The literature review revealed 479 papers and reports on the topic, from which 87 qualified for closer analysis. The review considered in- and out-of-plane patient shielding as well as shielding for pregnant and pediatric patients. Dose savings and other dose and non-dose related effects of patient shielding were considered in the evaluation. <h3>Conclusions</h3> Although patient shielding has been used in radiological practice for many years, its use is no longer undisputed. The evaluation of the systematic literature review of recent studies and reports shows that dose savings are rather minimal while significant dose- and non-dose-related detrimental effects are present. Consequently, the routine usage of patient protection shielding in medical X-ray imaging can be safely discontinued for all modalities and patient groups.

Deep Learning Based Classification of Wrist Cracks from X-ray Imaging

Wrist cracks are the most common sort of cracks with an excessive occurrence rate. For the routine detection of wrist cracks, conventional radiography (X-ray medical imaging) is used but periodically issues are presented by crack depiction. Wrist cracks often appear in the human arbitrary bone due to accidental injuries such as slipping. Indeed, many hospitals lack experienced clinicians to diagnose wrist cracks. Therefore, an automated system is required to reduce the burden on clinicians and identify cracks. In this study, we have designed a novel residual network-based convolutional neural network (CNN) for the crack detection of the wrist. For the classification of wrist cracks medical imaging, the diagnostics accuracy of the RN-21CNN model is compared with four well-known transfer learning (TL) models such as Inception V3, Vgg16, ResNet-50, and Vgg19, to assist the medical imaging technologist in identifying the cracks that occur due to wrist fractures. The RN-21CNN model achieved an accuracy of 0.97 which is much better than its competitor`s approaches. The results reveal that implementing a correct generalization that a computer-aided recognition system precisely designed for the assistance of clinician would limit the number of incorrect diagnoses and also saves a lot of time.

Full length image stitching algorithm for spinal deformity surgery

In the orthopedic surgery of the spine, obtaining a full-length spine image play an important role in evaluating the quality of the procedure. The most common imaging devices used to obtain spine images are C-arm x-ray machines, which requires multiple imaging of the patient's spine at different locations, then stitching together multiple images to obtain an entire spine. This paper proposes an approach based on an improved SIFT feature descriptor for the spine image stitching problem. Considering that the contrast of medical spine X-ray images is generally low and the contrast difference of individual images is significant, an improved SIFT algorithm is proposed based on image contrast information to adjust the contrast threshold. The improved SIFT was introduced into the stitching of spine X-ray images to achieve orthopedics spine images stitched quickly, accurately, and intra-operatively.

Multi-class Chest X-ray classification of Pneumonia, Tuberculosis and Normal X-ray images using ConvNets

Pneumonia and Tuberculosis (TB) are two serious and life-threatening diseases that are caused by a bacterial or viral infection of the lungs and have the potential to result in severe consequences within a short period of time. Therefore, early diagnosis is a significant factor in terms of a successful treatment process. Chest X-Rays which are used to diagnose Pneumonia and/or Tuberculosis need expert radiologists for evaluation. Thus, there is a need for an intelligent and automatic system that has the capability of diagnosing chest X-rays, and to simplify the disease detection process for experts and novices. This study aims to develop a model that will help with the classification of chest X-ray medical images into normal vs Pneumonia or Tuberculosis. Medical organizations take a minimum of one day to classify the diagnosis, while our model could perform the same classification within a few seconds. Also, it will display a prediction probability about the predicted class. The model had an accuracy, precision and recall score over 90% which indicates that the model was able to identify patterns. Users can upload their respective chest X-ray image and the model will classify the uploaded image into normal vs abnormal.

Does the Regulation of X-Ray Medical Imaging Need to Be Updated in the Pandemic Situation?

Does the Regulation of X-Ray Medical Imaging Need to Be Updated in the Pandemic Situation?

Swarm optimisation-based bag of visual words model for content-based X-ray scan retrieval

Classification and retrieval of medical images (MedIR) are emerging applications of computer vision for enabling intelligent medical diagnostics. Medical images are multi-dimensional and require specialised processing for the extraction of features from their manifold underlying content. Existing models often fail to consider the inherent characteristics of data and have thus often fallen short when applied to medical images. In this paper, we present a MedIR approach based on the bag of visual words (BoVW) model for content-based medical image retrieval. When it comes to any medical approach models, an imbalance in the dataset is one of the issues. Hence the perspective is also considering a balanced set of categories from an imbalanced dataset. The proposed work on BoVW model extracts features from each image are used to train supervised machine learning classifier for X-ray medical image classification and retrieval. During the experimental validation, the proposed model performed well with the classification accuracy of 89.73% and a good retrieval result using our filter-based approach.

Content Based Medical Image Retrieval for Accurate Disease Diagnosis

Introduction: Content Based Image Retrieval (CBIR) system is an innovative technology to retrieve images from various media types. One of the CBIR applications is Content Based Medical Image Retrieval (CBMIR). The image retrieval system retrieves the most similar images from the historical cases, and such systems can only support the physician's decision to diagnose a disease. To extract the useful features from the query image for linking similar types of images is the major challenge in the CBIR domain. The Convolution Neural Network (CNN) can overcome the drawbacks of traditional algorithms, dependent on the low-level feature extraction technique. Objective: The objective of the study is to develop a CNN model with a minimum number of convolution layers and to get the maximum possible accuracy for the CBMIR system. The minimum number of convolution layers reduces the number of mathematical operations and the time for the model's training. It also reduces the number of training parameters, like weights and bias. Thus, it reduces the memory requirement for the model storage. This work mainly focused on developing an optimized CNN model for the CBMIR system. Such systems can only support the physicians' decision to diagnose a disease from the images and retrieve the relevant cases to help the doctor decide the precise treatment. Methods: The deep learning-based model is proposed in this paper. The experiment is done with several convolution layers and various optimizers to get the maximum accuracy with a minimum number of convolution layers. Thus, the ten-layer CNN model is developed from scratch and used to derive the training and testing images' features and classify the test image. Once the image class is identified, the most relevant images are determined based on the Euclidean distance between the query features and database features of the identified class. Based on this distance, the most relevant images are displayed from the respective class of images. The general dataset CIFAR10, which has 60,000 images of 10 different classes, and the medical dataset IRMA, which has 2508 images of 9 various classes, have been used to analyze the proposed method. The proposed model is also applied for the medical x-ray image dataset of chest disease and compared with the other pre-trained models. Results: The accuracy and the average precision rate are the measurement parameters utilized to compare the proposed model with different machine learning techniques. The accuracy of the proposed model for the CIFAR10 dataset is 93.9%, which is better than the state-of-the-art methods. After the success for the general dataset, the model is also tested for the medical dataset. For the x-ray images of the IRMA dataset, it is 86.53%, which is better than the different pre-trained model results. The model is also tested for the other x-ray dataset, which is utilized to identify chest-related disease. The average precision rate for such a dataset is 97.25%. Also, the proposed model fulfills the major challenge of the semantic gap. The semantic gap of the proposed model for the chest disease dataset is 2.75%, and for the IRMA dataset, it is 13.47%. Also, only ten convolution layers are utilized in the proposed model, which is very small in number compared to the other pre-trained models. Conclusion: The proposed technique shows remarkable improvement in performance metrics over CNN-based state-of-the-art methods. It also offers a significant improvement in performance metrics over different pre-trained models for the two different medical x-ray image datasets.

Content Based Medical Image Retrieval for Accurate Disease Diagnosis

Introduction: Content Based Image Retrieval (CBIR) system is an innovative technology to retrieve images from various media types. One of the CBIR applications is Content Based Medical Image Retrieval (CBMIR). The image retrieval system retrieves the most similar images from the historical cases, and such systems can only support the physician's decision to diagnose a disease. To extract the useful features from the query image for linking similar types of images is the major challenge in the CBIR domain. The Convolution Neural Network (CNN) can overcome the drawbacks of traditional algorithms, dependent on the low-level feature extraction technique. Objective: The objective of the study is to develop a CNN model with a minimum number of convolution layers and to get the maximum possible accuracy for the CBMIR system. The minimum number of convolution layers reduces the number of mathematical operations and the time for the model's training. It also reduces the number of training parameters, like weights and bias. Thus, it reduces the memory requirement for the model storage. This work mainly focused on developing an optimized CNN model for the CBMIR system. Such systems can only support the physicians' decision to diagnose a disease from the images and retrieve the relevant cases to help the doctor decide the precise treatment. Methods: The deep learning-based model is proposed in this paper. The experiment is done with several convolution layers and various optimizers to get the maximum accuracy with a minimum number of convolution layers. Thus, the ten-layer CNN model is developed from scratch and used to derive the training and testing images' features and classify the test image. Once the image class is identified, the most relevant images are determined based on the Euclidean distance between the query features and database features of the identified class. Based on this distance, the most relevant images are displayed from the respective class of images. The general dataset CIFAR10, which has 60,000 images of 10 different classes, and the medical dataset IRMA, which has 2508 images of 9 various classes, have been used to analyze the proposed method. The proposed model is also applied for the medical x-ray image dataset of chest disease and compared with the other pre-trained models. Results: The accuracy and the average precision rate are the measurement parameters utilized to compare the proposed model with different machine learning techniques. The accuracy of the proposed model for the CIFAR10 dataset is 93.9%, which is better than the state-of-the-art methods. After the success for the general dataset, the model is also tested for the medical dataset. For the x-ray images of the IRMA dataset, it is 86.53%, which is better than the different pre-trained model results. The model is also tested for the other x-ray dataset, which is utilized to identify chest-related disease. The average precision rate for such a dataset is 97.25%. Also, the proposed model fulfills the major challenge of the semantic gap. The semantic gap of the proposed model for the chest disease dataset is 2.75%, and for the IRMA dataset, it is 13.47%. Also, only ten convolution layers are utilized in the proposed model, which is very small in number compared to the other pre-trained models. Conclusion: The proposed technique shows remarkable improvement in performance metrics over CNN-based state-of-the-art methods. It also offers a significant improvement in performance metrics over different pre-trained models for the two different medical x-ray image datasets.

High-Energy Electromagnetic Wave Radiation Analysis Study of Laboratories Using Digital Medical Imaging Devices

Devices that acquire digital medical images using radiation (DR; Digital Radiography) are widely used in medical institutions to obtain information necessary for patient diagnosis, and the frequency of use is gradually increasing. By selecting 3 hospitals that perform more than 50 digital X-ray medical imaging tests per day, and by installing a glass dosimeter for environmental measurement on the outside of the examination room protection wall, entrance door, and patient viewing window, the leakage radiation dose was measured for one month from May 16, 2021 to June 21, 2021, then it was analyzed by converting into a 3-month cumulative dose. The cumulative dose equivalent of 1 cm in the glass dosimeter for 3 months outside the shielding wall at the entrance of the digital medical image acquisition radiation generating device examination room was 0.96 mSv, and the average value was 0.18 mSv. In addition, the cumulative 1 cm dose equivalent value of the glass dosimeter for 3 months outside the shielding wall on the control room side showed a maximum of 3.03 mSv and an average of 0.41 mSv. The glass dosimeter cumulative dose 1cm dose equivalent value was up to 8.31 mSv and averaged 2.09 mSv for 3 months outside the door of the patient waiting room of the digital medical image acquisition radiation generating device examination room. In addition, the glass dosimeter cumulative dose 1cm dose equivalent value for 3 months outside the control room door showed a maximum of 14.04 mSv and an average of 3.84 mSv. These research results are expected to be widely used in setting measurement methods and regulatory standards for environmental radiation safety management of medical radiation generators, and continuous research management at the national level is suggested.

Noise Reduction in Medical X-Ray Images Using Wavelet and Neural Networks

Noise reduction in x-rays has attracted many researcher in recent years. This noise is produced duo to reduction of X-ray powers. Increasing the X-ray powers threatens the human life. Wavelet domain threshold level proposed by many researchers for this noise reduction. However, the many cases this approach is invalid because the noise distribution function is poison. In this research study we used the wavelet domain based on a genetic algorithm to tune the BayesShrink threshold. To increase the clarity an image, we used a multi-layer perceptron neural network. However, this neural network and other approaches are not able to eliminate large amount of noises. To overcome this kind of noise, we used a Directional adaptive median filter. Because the edges of images may be vanished by this method, we reconstructed the edge level at the end of process. The simulation results indicate that our approach provides more clear images than many of other methods in the literature such as PSNR, MSR and CNR.