Noise In X-ray Images Research Articles

Denoising method for X-ray images with poisson-Gaussian noise based on a new threshold function and shearlet transform

BackgroundX-ray imaging has been applied in various fields. However, the noise in X-ray images reduces the image quality and affects subsequent detection. AimsA denoising method is developed to solve the problem of the Poisson-Gaussian mixed noise caused by X-ray imaging. Methodology: A new threshold function and an improved threshold are proposed in this paper. Furthermore, an improved denoising method, namely the improved generalized Anscombe with shearlet transform (improved GA-ST), is developed based on the above proposed algorithms. After theoretical derivation, experiments and parameter analysis, the proposed method is applied to actual X-ray images. ResultsThe results show that the new threshold function is continuous, asymptotic, and has no inherent deviation, which solves the problems existing in traditional threshold functions. In addition, the improved GA-ST method can reduce Poisson-Gaussian mixed noise at different levels. As for actual X-ray images, the improved GA-ST method outperforms the other methods, and the BRISQUE descent ratios all exceed 25%. ConclusionsThe improved GA-ST method proposed in this paper can effectively reduce the noise in X-ray images and meet the requirements of actual applications based on MATLAB platform.

Removing Adversarial Noise in X-ray Images via Total Variation Minimizationand Patch-Based Regularization for Robust DeepLearning-based Diagnosis.

Deep learning has significantly advanced the field of radiology-based disease diagnosis, offering enhanced accuracy and efficiency in detecting various medical conditions through the analysis of complex medical images such as X-rays. This technology's ability to discern subtle patterns and anomalies has proven invaluable for swift and accurate disease identification. The relevance of deep learning in radiology has been particularly highlighted during the COVID-19 pandemic, where rapid and accurate diagnosis is crucial for effective treatment and containment. However, recent research has uncovered vulnerabilities in deep learning models when exposed to adversarial attacks, leading to incorrect predictions. In response to this critical challenge, we introduce a novel approach that leverages total variation minimization to combat adversarial noise within X-ray images effectively. Our focus narrows to COVID-19 diagnosis as a case study, where we initially construct a classification model through transfer learning designed to accurately classify lung X-ray images encompassing no pneumonia, COVID-19 pneumonia, and non-COVID pneumonia cases. Subsequently, we extensively evaluated the model's susceptibility to targeted and un-targetedadversarial attacksby employing the fast gradient sign gradient (FGSM) method. Our findings reveal a substantial reduction in the model's performance, with the average accuracy plummeting from 95.56 to 19.83% under adversarial conditions. However, the experimental results demonstrate the exceptional efficacy of the proposed denoising approach in enhancing the performance of diagnosis models when applied to adversarial examples. Post-denoising, the model exhibits a remarkable accuracy improvement, surging from 19.83 to 88.23% on adversarial images. These promising outcomes underscore the potential of denoising techniques to fortify the resilience and reliability of AI-based COVID-19 diagnostic systems, laying the foundation for their successful deployment in clinical settings.

An enhanced Convolutional Neural Network (CNN) model for the detection of lung cancer using X-Ray image

Lung Cancer is a life-threatening disease which can be diagnosed by Medical Imaging such as X-Ray, MRI, CT Scan etc. This research presented an enhanced model using Convolutional Neural Network (CNN) to detect lung cancer using X-Ray image. Medical image processing relies heavily on the diagnosis of lung cancer images. It aids doctors in determining the correct diagnosis and management. For many patients, lung cancer ranks among the mostdeadly diseases. Many lives can be saved if cancerous growth is diagnosed early. The purported model was predominantly built on Convolutional Neural Network (CNN) architecture and the model was built with enhanced features such as Image Enhancement, Segmenting ROI (Region of Interest), Features Extraction and Nodule Classification. In preprocessing stage, the AMF (Adaptive Median Filter) filtering method was applied to eliminate noise in X-Ray image of the dataset, and quality of X-Ray image was improved with the support of CLAHE (Contrast Limited Adaptive Histogram Equalization). Secondly, K-means Clustering algorithm was used to extract the relevant feature or Region of Interest (ROI) of the lung field automatically i.e. the model was effectively trained to identify and crop the exact location of the lung field automatically. The model was able to classify the cancer nodule as either Cancerous or Non-Cancerous. The framework worked on C# platform, and used EMGU for detection of the tumour in the lung xray image. Experimental result showed that the developed system was able to detect Lung Cancer with 90.77% accuracy, 86.65% precision and 95.31% Recall/Sensitivity.

Optimization Method to Predict Optimal Noise Reduction Parameters for the Non-Local Means Algorithm Based on the Scintillator Thickness in Radiography.

The resulting image obtained from an X-ray imaging system depends significantly on the characteristics of the detector. In particular, when an X-ray image is acquired by thinning the detector, a relatively large amount of noise inevitably occurs. In addition, when a thick detector is used to reduce noise in X-ray images, blurring increases and the ability to distinguish target areas deteriorates. In this study, we aimed to derive the optimal X-ray image quality by deriving the optimal noise reduction parameters based on the non-local means (NLM) algorithm. The detectors used were of two thicknesses (96 and 140 μm), and images were acquired based on the IEC 62220-1-1:2015 RQA-5 protocol. The optimal parameters were derived by calculating the edge preservation index and signal-to-noise ratio according to the sigma value of the NLM algorithm. As a result, a sigma value of the optimized NLM algorithm (0.01) was derived, and this algorithm was applied to a relatively thin X-ray detector system to obtain appropriate noise level and spatial resolution data. The no-reference-based blind/referenceless image spatial quality evaluator value, which analyzes the overall image quality, was best when using the proposed method. In conclusion, we propose an optimized NLM algorithm based on a new method that can overcome the noise amplification problem in thin X-ray detector systems and is expected to be applied in various photon imaging fields in the future.

Three-dimensional X-ray imaging of macropore flow

Macropores are known to be important pathways for the rapid transport of water, solutes and colloids in soil. Nevertheless, we still know very little about how the topology and geometry of macropore networks govern water flow configurations and velocities in natural soil. In this study, we aimed at gaining more insight into macropore flow by using X-ray tomography to quantify air–water distributions in the macropore networks of undisturbed topsoil and subsoil columns of a clay soil at varying steady-state flow rates. We observed that while large fractions of the macropore network remained air-filled, the air phase only became entrapped when the irrigation rate was very close to the saturated hydraulic conductivity of the soil. The data enabled us to parameterize a kinematic wave model for water flow following the approach proposed in Jarvis et al. (2017a). Follow-up experiments would be required to evaluate whether these kinematic wave parameters derived by X-ray imaging match with those obtained from outflow measurements. We found that quantitative X-ray imaging of macropore flow through soils still remains a challenging task. We recommend that future experiments are conducted on smaller soil samples to improve image resolution and minimize experimental time spans as well as X-ray image noise and illumination bias. Such experiments could also include 3-D tracer imaging to identify the imaged macropore networks transporting most of the water (i.e. the backbone) at varying steady irrigation rates.

Transfer Learning-Based Osteoporosis Classification Using Simple Radiographs

Osteoporosis is a condition that affects the entire skeletal system, resulting in decreased density of bone mass and the weakening of bone tissue's micro-architecture. This leads to weaker bones that are more susceptible to fractures. Detecting and measuring bone mineral density has always been a critical area of focus for researchers in the diagnosis of bone diseases such as osteoporosis. However, existing algorithms used for osteoporosis diagnosis encounter challenges in obtaining accurate results due to X-ray image noise and variations in bone shapes, especially in low contrast conditions. Therefore, the development of efficient algorithms that can mitigate these challenges and improve the accuracy of osteoporosis diagnosis is essential. In this research paper, a comparative analysis was conducted Assessing the accuracy and efficiency of the latest deep learning CNN model, such as VGG16, VGG19, DenseNet121, Resnet50 and InceptionV3 in detecting to Classify Normal and Osteoporosis cases. The study employed 830 X-ray images of Spine, Hand, Leg, Knee, and Hip, comprising of Normal (420) and Osteoporosis (410) cases. Various performance metrics were utilized to evaluate each model, and the findings indicate that DenseNet121 exhibited superior performance with an accuracy rate of 93.4% with Achieving an error rate of 0.07 and a validation loss of only 0.57 in comparison with other models considered in this study.

Switching median filter for suppressing multi-pixel impulse noise

This paper proposes a new switching median filter for suppressing multi-pixel impulse noise in X-ray images. A multi-pixel impulse is understood as a set of several neighboring pixels, the intensity of each significantly exceeds background intensity. Multi-pixel noise can occur, for example, due to the blooming effect, the reason being the limited value of pixel saturation capacity. This article defines the thresholds for the intensity increment relative to the eight immediate neighbors, above which the current pixel is processed by the median filter. The dependence of these thresholds on the number of pixels in an impulse is presented. The proposed algorithm is based on the median filtering process, which consists of several iterations. In this case, the filter has the smallest possible size, which minimizes image distortion during processing. In particular, to exclude a single-pixel impulse, pixel processing is turned on when intensity surge exceeds 3.5 with the grayscale value ranging from 0 to 1. At the same time, to exclude nine-pixel impulses, three iterations are required with the following thresholds: the first iteration with a threshold 2.0; the second iteration also with a threshold 2.0 and the third iteration with a threshold 3.5. The algorithm proposed was tested on real X-ray images corrupted by multi-pixel impulse noise. The algorithm is not only simple, but also reliable and suitable for real-time implementation and application. The efficiency of the technique is shown in comparison with other known filtering methods with respect to the degree of noise suppression. The main result of the testing is that only the proposed method allows excluding multi-pixel noise. Other advantage of the algorithm is its weak effect on the level of Gaussian noise leading to the absence of image blurring (or preserving image details) during processing.

Poisson-Gaussian Noise Reduction for X-Ray Images Based on Local Linear Minimum Mean Square Error Shrinkage in Nonsubsampled Contourlet Transform Domain

Noise reduction is important for X-ray images because it can reduce radiation exposure to patients. X-ray image noise has a Poisson-Gaussian distribution, and recently, noise analysis and removal in multiscale transformations have been widely implemented. The nonsubsampled contourlet transform (NSCT) is a multiscale transformation suitable for medical images that separates the scale and direction. This study proposes a Poisson-Gaussian noise-removal method using NSCT shrinkage that is based on the characteristics of Poisson-Gaussian noise in NSCT domain. It has the structure of a block-matching 3D filtering algorithm in the form of basic estimation and noise removal process; however, the main processes are modified to consider Poisson-Gaussian noise characteristics. In the basic estimation process, an NSCT shrinkage method that is suitable for Poisson-Gaussian noise characteristics is developed by optimizing the local linear minimum mean square error estimator in the NSCT domain. In the denoising step, the noise term of the Wiener filter is determined using the result of the NSCT shrinkage, and finally, the denoised image is obtained. The proposed method is applied to simulated and real X-ray images and is compared with other state-of-the-art Poisson-Gaussian noise removal methods; it exhibits excellent results in both quantitative and qualitative aspects.

Effectiveness of Non-Local Means Algorithm with an Industrial 3 MeV LINAC High-Energy X-Ray System for Non-Destructive Testing.

Industrial high-energy X-ray imaging systems are widely used for non-destructive testing (NDT) to detect defects in the internal structure of objects. Research on X-ray image noise reduction techniques using image processing has been widely conducted with the aim of improving the detection of defects in objects. In this paper, we propose a non-local means (NLM) denoising algorithm to improve the quality of images obtained using an industrial 3 MeV high-energy X-ray imaging system. We acquired X-ray images using various castings and assessed the performance visually and by obtaining the intensity profile, contrast-to-noise ratio, coefficient of variation, and normalized noise power spectrum. Overall, the quality of images processed by the proposed NLM algorithm is superior to those processed by existing algorithms for the acquired casting images. In conclusion, the NLM denoising algorithm offers an efficient and competitive approach to overcome the noise problem in high-energy X-ray imaging systems, and we expect the accompanying image processing software to facilitate and improve image restoration.

Diagnostic and Gradation Model of Osteoporosis Based on Improved Deep U-Net Network.

The measurement of bone mineral density for osteoporosis has always been the focus of researchers because it plays an important role in bone disease diagnosis. However, because of X-ray image noise and the large difference between the bone shapes of patients under the condition of low contrast, existing osteoporosis diagnosis algorithms are difficult to obtain satisfactory results. This paper presents an improved osteoporosis diagnosis algorithm based on U-NET network. Firstly, the bone in the original image are marked and used to construct the data set. And then, by normalizing the input of each layer, it can be ensured that the input data distribution of each layer is stable, so that the purpose of accelerated training can be achieved. Finally, the energy function is calculated by combining the value of the softmax prediction class for each pixel on the final feature map with the Cross entropy loss function and all the segmented images are extracted to obtain the diagnostic result. As the experimental results show that the improved U-net can accurately solve the influence of image interference in the process of bone mineral density measurement. The recognition rate of U-net automatic diagnosis method is above 81%, and the diagnosis effect is better than other comparison methods.

Feasibility of newly designed fast non local means (FNLM)-based noise reduction filter for X-ray imaging: A simulation study

In the diagnostic radiology field, reducing the radiation dose for patient lead to increased noise in image. Since increases of noise decrease the diagnosis rate, to reduce the noise is necessary. In this study quantitatively evaluates the four widely used and newly verified filters which remove noise in image: median, Wiener, total variation, and fast non local means (FNLM). For that purpose, X-ray and computed tomography (CT) images are acquired using MATLAB simulation with 3D voxelized phantom. To evaluate image performance, normalized noise power spectrum (NNPS), contrast to noise ratio (CNR) and coefficient of variation (COV) were used. As a result, we can efficiently remove noise in X-ray image when FNLM filter was used compared with frequently used filters. In conclusion, our results demonstrated that our proposed FNLM filter shows superior denoising performance, which is expected to enhance the detection of diseases in clinical images with low dose.

Novel X-ray image noise reduction technology reduces patient radiation dose while maintaining image quality in coronary angiography.

AimsThe consequences of high radiation dose for patient and staff demand constant improvements in X-ray dose reduction technology. This study assessed non-inferiority of image quality and quantified patient dose reduction in interventional cardiology for an anatomy-specific optimised cine acquisition chain combined with advanced real-time image noise reduction algorithms referred to as ‘study cine’, compared with conventional angiography.MethodsFifty patients underwent two coronary angiographic acquisitions: one with advanced image processing and optimised exposure system settings to enable dose reduction (study cine) and one with standard image processing and exposure settings (reference cine). The image sets of 39 patients (18 females, 21 males) were rated by six experienced independent reviewers, blinded to the patient and image characteristics. The image pairs were randomly presented. Overall 85 % of the study cine images were rated as better or equal quality compared with the reference cine (95 % CI 0.81–0.90). The median dose area product per frame decreased from 55 to 26 mGy.cm2/frame (53 % reduction, p < 0.001).ConclusionThis study demonstrates that the novel X-ray imaging technology provides non-inferior image quality compared with conventional angiographic systems for interventional cardiology with a 53 % patient dose reduction.

Extracting and Mathematical Identifying Form of Stationary Noise in X-ray Images

Image noise may prevent proper diagnostic X-ray imaging. This study is aimed at developing new noise rejection methods using a mathematical model that describes the form of X-ray image noise. Stationary noise is one type of noise found in X-ray images. Stationary noise is nonstochastic and appears independent of the radiographic factors. In this paper, we verify methods for identifying stationary noise using a polynomial regression model, and extracting such noise from X-ray images obtained from a CR system. The results of this study demonstrate that stationary noise can be extracted with high precision using a particular low-pass filter frequency. We found that a regression model for greater than second-degree polynomials can be applied for roughly identifying stationary noise. However, the fitting accuracy of the regression curve is not significantly improved in terms of the amount of multiplication required when increasing the degree of the polynomial regression model. Ke yw ords: X-ray Image, Nonstochastic Noise, Stationary Noise, Polynomial Regression Model and CR System

Continuous Shape Estimation of Continuum Robots Using X-ray Images.

We present a new method for continuously and accurately estimating the shape of a continuum robot during a medical procedure using a small number of X-ray projection images (e.g., radiographs or fluoroscopy images). Continuum robots have curvilinear structure, enabling them to maneuver through constrained spaces by bending around obstacles. Accurately estimating the robot's shape continuously over time is crucial for the success of procedures that require avoidance of anatomical obstacles and sensitive tissues. Online shape estimation of a continuum robot is complicated by uncertainty in its kinematic model, movement of the robot during the procedure, noise in X-ray images, and the clinical need to minimize the number of X-ray images acquired. Our new method integrates kinematics models of the robot with data extracted from an optimally selected set of X-ray projection images. Our method represents the shape of the continuum robot over time as a deformable surface which can be described as a linear combination of time and space basis functions. We take advantage of probabilistic priors and numeric optimization to select optimal camera configurations, thus minimizing the expected shape estimation error. We evaluate our method using simulated concentric tube robot procedures and demonstrate that obtaining between 3 and 10 images from viewpoints selected by our method enables online shape estimation with errors significantly lower than using the kinematic model alone or using randomly spaced viewpoints.

Advanced template matching method for estimation of intervertebral kinematics of lumbar spine

Diagnosis of low back pain and other degenerative spinal pathologies can be extremely difficult and, so far, there are not accepted standards. In general, such pathologies are associated with alteration of mechanical properties of spine and, in particular, with the instability of spinal motion. Intervertebral kinematics can be a valuable, objective method to assess the functionality of spinal segments. Fluoroscopic imaging system can provide continuous screening of lumbar tracts during patient's motion, with an acceptable low X-ray dose. Estimation of intervertebral kinematics relies on accurate recognition of vertebrae positions throughout the fluoroscopic sequence: specific vertebrae features are identified and tracked either by manual selection or by automated methods. This study presents a new method of vertebra tracking, based on image template matching of the contour of the vertebral body for an accurate intervertebral kinematics analysis. An image gradient operator was utilized to obtain the vertebral contours; it operates after an edge-preserving smoothing filter designed to reduce low dose X-ray image noise. Once a template is defined for each vertebra, this is used to determine the best vertebral location in each image throughout the fluoroscopic sequence. Accuracy of the proposed method was tested using images of a calibration model. Average error achieved for the intervertebral angle is of the order of 0.4° and approximately 2 mm for the intervertebral centre of rotation. Five fluoroscopic lumbar sequences of healthy volunteers undergoing passive flexion–extension motion were processed. The intervertebral kinematics was compared with other methods (automated and manual) by an estimation of measurement error. Results showed that the current method provides a better representation of the evolution over time of kinematic parameters. In particular, root mean square differences between the current method and a manual selection procedure performed by an experienced and trained clinician resulted 1.3° for the intervertebral angles and 0.9 mm for the intervertebral trajectory. The proposed method provides an effective, automated and objective technique for estimation of intervertebral kinematics of lumbar spine.

Method for reducing noise in X-ray images by averaging pixels based on the normalized difference with the relevant pixel

A real-time digital filter for noise reduction in X-ray images is proposed. The filter is based on averaging of only similar pixels (pixels that differ only little) rather than neighboring pixels, which are averaged in conventional linear low-pass filters. The effectiveness of the filter was evaluated by computer simulation, where original images that were acquired by X-ray exposure were processed in accordance with the filter algorithm. The resulting images were evaluated in terms of the pre-sampled modulation transfer function (MTF), the noise power spectrum (NPS), and the lag. Comparison of the filtered and original images revealed that the NPS was reduced for the full range of spatial frequencies in the filtered image, resulting in a reduction of total noise power to about 1/9 the level in the original image with no degradation in the MTF or lag. The usefulness of the filter was demonstrated in fluoroscopic, digital subtraction angiography (DSA) and mammographic phantom studies. The filter was found to have the potential to reduce the patient dose by reducing the noise in dynamic as well as static X-ray images.

ROC Study of Four LCD Displays Under Typical Medical Center Lighting Conditions

Nine observers reviewed a previously assembled library of 320 chest computed radiography (CR) images. Observers participated in four sessions, reading a different 1/4 of the sample on each of four liquid crystal displays: a 2-megapixel (MP) consumer color display, a 2-MP business color display, a 2-MP medical-grade gray display, and a 3-MP gray display. Each display was calibrated according to the DICOM Part 14 standard. The viewing application required observer log-in, then randomized the order of the subsample seen on the display, and timed the responses of the observer to render a 1-5 judgment on the absence or presence of ILD on chest CRs. Selections of 1-2 were considered negative, 3 was indeterminate, and 4-5 were positive. The order of viewing sessions was also randomized for each observer. The experiment was conducted under controlled lighting, temperature, and sound conditions to mimic conditions typically found in a patient examination room. Lighting was indirect, and illuminance at the display face was 195 +/- 8% lux and was monitored over the course of the experiment. The average observer sensitivity for the 2 MP color consumer, 2 MP business color, 2 MP gray, and 3 MP gray displays were 83.7%, 84.1%, 85.5%, and 86.7%, respectively. The only pairwise significant difference was between the 2-MP consumer color and the 2-MP gray (P = 0.05). Effect of order within a session was not significant (P = 0.21): period 1 (84.3%), period 2 (86.2%), period 3 (85.4%), period 4 (84.1%). Observer specificity for the various displays was not statistically significant (P = 0.21). Finally, a timing analysis showed no significant difference between the displays for the user group (P = 0.13), ranging from 5.3 s (2 MP color business) to 5.9 s (3 MP Gray). There was, however, a reduction in time over the study that was significant (P < 0.001) for all users; the group average decreased from 6.5 to 4.7 s per image. Physical measurements of the resolution, contrast, and noise properties of the displays were acquired. Most notably, the noise of the displays varied by 3.5x between the lowest and highest noise displays. Differences in display noise were indicative of observer performance. However, the large difference in the magnitude of the noise was not predictive of the small difference (3%) in the observer sensitivity for various displays. This is likely because detection of interstitial lung disease is limited by "anatomical noise" rather than display or x-ray image noise.