Intensity Inhomogeneity Research Articles

Effects of Dementia and MCI on Diffusion Tensor Metrics Using the Updated ADNI3 DTI Preprocessing Pipeline

AbstractBackgroundDiffusion magnetic resonance imaging (dMRI) provides insight into white matter (WM) microstructural changes in Alzheimer’s disease and mild cognitive impairment (MCI). The Alzheimer’s Disease Neuroimaging Initiative’s (ADNI) has currently extended its ADNI3 imaging protocol to include dMRI from Siemens, Philips and General Electric (GE) scanners, resulting in seven vendor‐specific dMRI acquisition protocols (Table1). Here we present the updated ADNI3 preprocessing pipeline1 (Figure1) for diffusion tensor imaging (DTI) metrics made available via ADNI’s database, and evaluate their utility for detecting WM differences associated with clinical impairment.MethodRaw dMRI downloaded from ADNI are first denoised using DiPy’s principal component analysis (PCA) denoising algorithms: local PCA for zero‐padded k‐space data (GE), or Marchenko‐Pastur PCA for data with the original acquisition matrix (Siemens/Philips). dMRI are then corrected for Gibbs ringing, eddy currents using FSL’s eddy_cuda with repol outlier estimation and slice‐to‐volume correction, intensity inhomogeneity with ANTS N4, and echo‐planar imaging distortions with a three‐channel non‐linear registration of the subject‐level preprocessed average B0 image to the subject’s skull‐stripped T1‐weighted MRI. Bias field correction is applied again to dMRI with FSL‐FAST. DTI fractional anisotropy (FA), axial, mean and radial diffusivity (AxD, MD, RD) are estimated with FSL’s dtifit, using weighted least squares. The JHU ICBM‐DTI‐81 atlas FA is warped to subject‐level FA maps using ANTS, and the transformations are applied to atlas labels. Subject‐level mean and robust mean (using M‐estimator from R’s ‘WRS2’ package) of DTI metrics within 73 WM regions of interest (ROIs; Table2) are extracted. Here, we fit linear mixed effects models of diagnosis (N=733) and clinical dementia rating (CDR‐sob; N=682; Table3) in 28 of the 73 available ROIs, covarying for age, sex, age*sex, and including nested protocol|site variables as random effects; multiple comparisons were corrected for using false discovery rate (q=0.05).ResultLower anisotropy and greater diffusivity were associated with greater impairment, particularly in the hippocampal cingulum (Figure2). Results from mean and robust mean metrics were comparable. Overall, our results follow previous work2,3 using publicly available ADNI dMRI.ConclusionSubject‐level ADNI3 dMRI measures, sensitive to clinical indices of impairment as presented here, are available at https://ida.loni.usc.edu/.

An Active Contour Model Based on Retinex and Pre-Fitting Reflectance for Fast Image Segmentation

In the present article, this paper provides a method for fast image segmentation for computer vision, which is based on a level set method. One dominating challenge in image segmentation is uneven illumination and inhomogeneous intensity, which are caused by the position of a light source or convex surface. This paper proposes a variational model based on the Retinex theory. To be specific, firstly, this paper figures out the pre-fitting reflectance by using an algorithm in the whole image domain before iterations; secondly, it reconstructs the image domain using an additive model; thirdly, it uses the deviation between the global domain and low-frequency component to approximate the reflectance, which is the significant part of an energy function. In addition, a new regularization term has been put forward to extract the vanishing gradients. Furthermore, the new regularization term is capable of accelerating the segmentation process. Symmetry plays an essential role in constructing the energy function and figuring out the gradient flow of the level set.

A novel approach for MR brain tumor classification and detection using optimal CNN‐SVM model

AbstractThe segmentation of brain tumors in magnetic resonance imaging plays a significant role in the field of image processing. This process has high computational complexity when handled manually by clinical experts. The accuracy in classifying and segmenting the brain tumor depends on the radiologists' experience. The computer‐aided diagnosis‐based brain tumor segmentation approach is proposed to overcome the existing limitations. The proposed convolutional neural network and support vector machine approach consists of the following stages. In the preprocessing stage, unwanted noise and intensity inhomogeneity are suppressed using an anisotropic diffusion filter. Then, the features are extracted using the deep convolutional neural network, and based on the features; the input brain image is classified into normal or abnormal using a support vector machine classifier. The proposed method gives a more successful accuracy rate of 2.11%. Compared with the other methods, the sensitivity and specificity values are also improved to 4.79% and 1.19%.

Cross-beam energy transfer between spatially smoothed laser beams

The crossing of two spatially smoothed laser beams amounts to the crossings of a large number of speckles. The energy transfer between two of these speckles is mediated by laser induced electron/ion density ripples that act as a Bragg grating. In a weakly Landau-damped plasma, this ion acoustic wave (IAW) may propagate from one crossing region to another, hence perturbing the local electron/ion grating [Oudin et al. Phys. Rev. Lett. 127, 265001 (2021)] even without phase shift between IAWs. In this paper, we investigate how the phase-shifted IAWs generated at the speckle scale interfere and affect the overall energy exchange. To this aim, we perform 2D particle-in-cell simulations with in-phase and out-of-phase Gaussian beams. In the latter situation, which better matches a smoothed laser beam, we find that the destructive interferences between the ion waves significantly reduce the energy exchange compared to the plane wave case. Additional 2D particle-in-cell simulations with random phase plate smoothed laser beams confirm the relevance of this effect in carbon plasma. A second effect is that cross-beam energy transfer (CBET) inhibition persists in strongly damped plasmas when the speckle radius is comparable with the IAW damping distance. There, the reduction in the IAW amplitude is attributed to the smallness of the speckle's envelop. These results are supported by a simple model that analytically estimates the CBET and clearly shows that neglecting the inhomogeneities in the laser intensity would usually lead to an overestimate of the energy exchange.

Detection of skin defects on loquat using hyperspectral imaging combining both band radio and improved three-phase level set segmentation method

AbstractBackground and objectivesSkin defects are one of the primary problems that occur in post-harvest grading and processing of loquats. Skin defects lead to the loquat being easily destroyed during transportation and storage, which causes the risk of other loquats being infected, affecting the selling price.Materials and MethodsIn this paper, a method combining band radio image with an improved three-phase level set segmentation algorithm (ITPLSSM) is proposed to achieve high accuracy, rapid, and non-destructive detection of skin defects of loquats. Principal component analysis (PCA) was used to find the characteristic wavelength and PC images to distinguish four types of skin defects. The best band ratio image based on characteristic wavelength was determined.ResultsThe band ratio image (Q782/944) based on PC2 image is the best segmented image. Based on pseudo-color image enhancement, morphological processing, and local clustering criteria, the band ratio image (Q782/944) has better contrast between defective and normal areas in loquat. Finally, the ITPLSSM was used to segment the processing band ratio image (Q782/944), with an accuracy of 95.28%.ConclusionsThe proposed ITPLSSM method is effective in distinguishing four types of skin defects. Meanwhile, it also effectively segments images with intensity inhomogeneities.

Optimized biparametric magnetic resonance imaging protocol for prostate cancer detection

BACKGROUND: Prostate cancer is one of the most commonly diagnosed cancers in men worldwide. PI-RADS v2.1 contains the requirements for the magnetic resonance imaging protocol, which cannot be fully implemented on a significant component of functioning scanners. Consequently, magnetic resonance imaging approaches vary in different medical organizations and often do not allow for a qualitative interpretation of images and diagnosis of the target pathology.
 AIM: To develop a biparametric magnetic resonance imaging protocol optimized for the existing magnetic resonance imaging scanners for the diagnosis of prostate cancer and to allow the screening and detection of neoplasms as early as possible. Simultaneously, the protocol should fulfill the current PI-RADS v2.1 recommendations to the maximum possible extent and meet the requirements of effective workflow in the radiology department.
 MATERIALS AND METHODS: Preliminary analysis of prostate magnetic resonance imaging scanning in medical organizations of the Moscow Health Care Department showed the absence of a unified approach. Using the iterative adjustment of scanning parameters, we adjusted the protocol to ensure acceptable quality with maximum available compliance with PI-RADS v2.1.
 To quantify the quality of the images, we used the magnetic resonance imaging phantom recommended by the American College of Radiology.
 RESULTS: The biparametric protocol was developed for Excelart Vantage 1.5 T, including T2-weighted images in three planes and diffusion-weighted images, which took less than 11 min. Moreover, the image quality parameters (intensity inhomogeneity, nonlinearity, resolution, and slice thickness) were within the acceptable ranges recommended by the magnetic resonance imaging manufacturer.
 CONCLUSION: The prostate may be effectively evaluated using the proposed magnetic resonance imaging protocol. Introducing it into practice could have a significant impact on the detection of prostate cancer in men. The entire duration of the protocol provides a possibility to supplement it with any sequences, depending on the final purpose of investigation.

The effect of temperature on the electrical characteristics of Ti/n-GaAs Schottky diodes

Schottky diodes still attract researchers as they are used in various device applications. This study provides I–V characteristics of Ti/n-GaAs (80–300 K). Higher barrier height (ΦB0) values were obtained for higher temperatures, whereas the ideality factor exhibited the opposite behavior. This was associated with a barrier inhomogeneity at the Ti/GaAs interface, which has a Gaussian distribution (GD). The mean barrier height values calculated from the modified Richardson and ΦB0 - q/2 kT plots were found to be 0.584 eV and 0.575 eV in the temperature range of 80–160 K. They were found as 1.041 eV and 1.033 eV between 180 K and 300 K, respectively. The modified Richardson constant value, on the other hand, was calculated as 22.06 A cm−2 K−2 (80–160 K) and 13.167 A cm−2 K−2 (180–300 K). These values are higher than the theoretical value for n-GaAs, which is 8.16 A cm−2 K−2. This difference may stem from intense inhomogeneity at the Ti/n-GaAs interface.

New foreground markers for Drosophila cell segmentation using marker-controlled watershed

Image segmentation consists of partitioning the image into different objects of interest. For a biological image, the segmentation step is important to understand the biological process. However, it is a challenging task due to the presence of different dimensions for cells, intensity inhomogeneity, and clustered cells. The marker-controlled watershed (MCW) is proposed for segmentation, outperforming the classical watershed. Besides, the choice of markers for this algorithm is important and impacts the results. For this work, two foreground markers are proposed: kernels, constructed with the software Fiji and Obj.MPP markers, constructed with the framework Obj.MPP. The new proposed algorithms are compared to the basic MCW. Furthermore, we prove that Obj.MPP markers are better than kernels. Indeed, the Obj.MPP framework takes into account cell properties such as shape, radiometry, and local contrast. Segmentation results, using new markers and illustrated on real Drosophila dataset, confirm the good performance quality in terms of quantitative and qualitative evaluation.

Differentiation between atypical sinonasal non-Hodgkin's lymphoma and inverted papilloma

To seek additional magnetic resonance imaging (MRI) features to improve the accuracy of differentiation between atypical sinonasal non-Hodgkin's lymphoma (NHL) and inverted papilloma (IP) using conventional MRI and apparent diffusion coefficient (ADC) maps. MRI examinations from 44 atypical cases (21 NHLs and 23 IPs) in sinonasal regions were reviewed retrospectively. Imaging features included tumour laterality, extension, T1-weighted imaging (WI)/T2WI signal intensity homogeneity and ratios, enhancement homogeneity and ratios, and ADCmean. In cases of NHL, homogeneous signal intensity was often observed on T2WI, which was homogeneous and significantly less enhanced than the turbinate, with lower ADCmean. Whereas in IPs, heterogeneous signal intensity was seen on T2WI, which was heterogeneous and of comparable enhancement to the turbinate, and higher ADCmean values were commonly seen. An ADCmean cut-off point of 1.10×10-3 mm2/s achieved 100% sensitivity, 90% specificity, and 90% accuracy. In addition, special features were observed that support the distinction between the two tumours, including intestinal pattern enhancement in NHL and spot-like appearance on T2WI and enhancement in IP. ADCmean was the most valuable metric for differentiating between the atypical sinonasal NHLs and IPs.

Detection of Road Cracks Using Convolutional Neural Networks and Threshold Segmentation

Automatic road crack detection is an important transportation maintenance responsibility for ensuring driving comfort and safety. Manual inspection is considered to be a risky method because it is time consuming, costly, and dangerous for the inspectors. Automated road crack detecting techniques have been extensively researched and developed in order to overcome this issue. Despite the difficulties, most of the proposed methodologies and solutions involve machine vision and machine learning, which have lately acquired traction largely due to the increasingly more affordable processing power. Nonetheless, it remains a difficult task due to the inhomogeneity of crack intensity and the intricacy of the background. In this paper, a convolutional neural network-based method for crack detection is proposed. The method is inspired from recent advancements in applying machine learning to computer vision. The primary goal of this work is to employ convolutional neural networks to detect the road crack. Data in the form of images has been used as input, preprocessing and threshold segmentation is applied to the input data. The processed output is feed to CNN for feature extraction and classification. The training accuracy was found to be 96.20 %, the validation accuracy to be 96.50 %, and the testing accuracy to be 94.5 %.

Segmentation of mammogram images using level set with Cuckoo search optimisation

ABSTRACT Breast cancer is alone accounted for 30% of all new cancer diagnoses for women. Early identification of breast cancer is a key stage in the process of diagnosis and treatment. Segmenting of low contrast mammograms is not an easy task due to irregular mass form, speculated margins and variation in intensity levels. Most of the region-based technique works well for homogenous images but fails for intensity inhomogeneity. Multiphase level set segmentation is optimised by the cuckoo search approach for energy minimisation. The proposed multiphase level set with the cuckoo optimisation method performs well in segmenting homogeneous as well as inhomogeneous regions. The Chan Vese (CV) and the proposed method are applied to the Mammographic Image Analysis Society (MIAS) database to obtain the accurate contour of the masses. To validate the proposed approach using quantitative performance measures such as Jaccard coefficient (JC), dice coefficient (DC), Hausdorff distance (Hd), sensitivity, specificity and accuracy were used for the MIAS data set. The results indicate that the suggested method can successfully segment the breast mass in mammograms and more robust than the classical CV method.

Jointly estimating bias field and reconstructing uniform MRI image by deep learning

Bias field is one of the main artifacts that degrade the quality of magnetic resonance images. It introduces intensity inhomogeneity and affects image analysis such as segmentation. In this work, we proposed a deep learning approach to jointly estimate bias field and reconstruct uniform image. By modeling the quality degradation process as the product of a spatially varying field and a uniform image, the network was trained on 800 images with true bias fields from 12 healthy subjects. A network structure of bias field estimation and uniform image reconstruction was designed to compensate for the intensity loss. To further evaluate the benefit of bias field correction, a quantitative analysis was made on image segmentation. Experimental results show that the proposed BFCNet improves the image uniformity by 8.3% and 10.1%, the segmentation accuracy by 4.1% and 6.8% on white and grey matter in T2-weighted brain images. Moreover, BFCNet outperforms the state-of-the-art traditional methods and deep learning methods on estimating bias field and preserving image structure, and BFCNet is robust to different levels of bias field and noise.

Efficient Echocardiographic Image Segmentation

In this paper, we propose an improved region-based active contour method based on the development of a novel signed pressure force (SPF) function. To obtain the required boundary, the method is applied to the echocardiographic images. Ultrasound image segmentation is particularly challenging due to speckle noise, low contrast, and intensity inhomogeneity. Because of these factors, segmenting echocardiographic images is a difficult task. All of these issues are addressed by the proposed model, which detects the true boundary without any noise. The proposed model is more robust, effective, and accurate when applied to images with weak edges and inhomogeneous intensity.

A Discriminative Level Set Method with Deep Supervision for Breast Tumor Segmentation

Background:Breast tumor segmentation in B-mode ultrasound imaging is important for analyzing, identifying, and diagnosing tumors. The level set is an approach most widely used in breast segmentation, and the refinement is still in progress. However, its effectiveness is harmed by a dearth of semantic information. On the other hand, deep networks contain rich semantic information but loss much influential low-level details. Method:This paper proposes a novel deep-feature embedded level set group to exploit semantically enriched features for breast tumor segmentation. First, a UNet-based network is trained to extract different features at different stages. Each stage has unique features depiction. Then, a novel level-set method is integrated at the end of each stage to approach more accurate and precise features maps. A new feature-discriminator is devised in the energy function of the level set method to refine the low confidence pixels at the boundaries. Lastly, the outputs of the level set method at different stages are incorporated into final feature maps to further empower the segmentation process. Two datasets comprising 349 breast ultrasound images from various hospitals have been utilized to assess the proposed approach’s performance. The model’s effectiveness is estimated on different metrics, including Accuracy, Sensitivity or True Positive rate, Specificity or True Negative rate, False Positive rate Dice, and IoU values for both datasets. Furthermore, the efficiency of the model is investigated by performing a comparison with several state-of-the-art classic segmentation methods and deep learning methods. Result:The proposed method outperformed segmenting breast ultrasound tumors in terms of Dice and IoU for datasets A and B (with p-value < 0.005 against compared methods). Additionally, the performance of the proposed approach is evaluated using the Area Under Receiver Operating Characteristics curve (AUC) and Mean Absolute Error (MAE). Our findings indicate that the proposed method seems to gain superiority over other methods by obtaining a lower MAE rate with the highest value of the AUC. Conclusion:Experiments determine that our method has obtained the best cut-off to deal with the noticeable glitches present in other approaches and generates more accurate segmentation results for tumors in complex images. Hence, the results confirm the proposed method’s effectiveness compared to classic segmentation methods over ultrasound images with blurry boundaries, noise, and intensity inhomogeneity. Moreover, our approach gives unprecedented prediction accuracy and similarity for malignant tumors.

Polynomial chaos expansion for uncertainty propagation analysis in numerical homogenization of 2D/3D periodic composite microstructures

This paper proposes the use of adaptive polynomial chaos expansion (PCE) for uncertainty propagation analysis of the numerical homogenization of polymer composites doped with random dispersions of spherical inclusions. The developed PCE acts as a surrogate model bypassing the computationally intense numerical homogenization of the elastic properties of 2D/3D representative volume elements (RVEs) loaded with moderate to high filler contents. Numerical results and discussion are presented to assess the accuracy and computational efficiency of 2D and 3D homogenization meta-models. The ability of the developed approaches to perform uncertainty propagation analyses with minimum computational effort represents the main contribution of this work, which holds vast potential for the stochastic design of macroscopic composite structural elements.

Adjacent Frame Difference with Dynamic Threshold Method in Underwater Flash Imaging LiDAR

During the underwater LiDAR imaging process, the images achieved by the conventional constant threshold adjacent frame difference (AFD) method normally loses the distance information of targets. This is mainly due to the Gaussian distribution of the laser light intensity field, which leads to the inhomogeneous intensity distribution in the frame from the target acquired by intensity charge-coupled devices (ICCD). In order to overcome this issue, the novel dynamic threshold adjacent frame difference (DTAFD) method was proposed in this paper. The DTAFD method modifies the intensity threshold following the pixel intensities in the different parts of the single frame intensity image acquired by ICCD. After the detailed theoretical demonstration of the DTAFD method, with the purpose of verifying its feasibility, the self-developed range-gated flash imaging LiDAR has been employed to construct the distance images of the rectangular and circular shaped targets at different distances. The distance between the rectangular target and the LiDAR system is 25.7 m, and the circular target is 70 cm further away from the rectangular target. The full distance information of these two targets is obtained by the DTAFD method with an effectively suppressing noise and the PSNR is increased from 6.95±0.0426 dB to 7.62±0.0264 dB. The experimental results indicate that the DTAFD method efficiently solves the AFD method’s drawback on the target information loss caused by the unequal optical field distribution, which makes it more suitable for the scenarios with uneven laser distribution such as the underwater imaging environment.

Image Segmentation with Adaptive Spatial Priors from Joint Registration

Image segmentation is a crucial but challenging task that has many applications. In medical imaging, for instance, intensity inhomogeneity and noise are common. In thigh muscle images, different muscles are closely packed together and there are often no clear boundaries between them. Intensity based segmentation models cannot separate one muscle from another. To solve such problems, in this work we present a segmentation model with adaptive spatial priors from joint registration. This model combines segmentation and registration in a unified framework to leverage their positive mutual influence. The segmentation is based on a modified Gaussian mixture model, which integrates intensity inhomogeneity and spatial smoothness. The registration plays the role of providing a shape prior. We adopt a modified sum of squared difference fidelity term and Tikhonov regularity term for registration and also utilize a Gaussian pyramid and parametric method for robustness. The connection between segmentation and registration is guaranteed by the cross entropy metric that aims to make the segmentation map (from segmentation) and deformed atlas (from registration) as similar as possible. This joint framework is implemented within a constraint optimization framework, which leads to an efficient algorithm. We evaluate our proposed model on synthetic and thigh muscle MR images. Numerical results show the improvement as compared to segmentation and registration performed separately and other joint models.

Multimodal image synthesis based on disentanglement representations of anatomical and modality specific features, learned using uncooperative relativistic GAN.

Growing number of methods for attenuation-coefficient map estimation from magnetic resonance (MR) images have recently been proposed because of the increasing interest in MR-guided radiotherapy and the introduction of positron emission tomography (PET) MR hybrid systems. We propose a deep-network ensemble incorporating stochastic-binary-anatomical encoders and imaging-modality variational autoencoders, to disentangle image-latent spaces into a space of modality-invariant anatomical features and spaces of modality attributes. The ensemble integrates modality-modulated decoders to normalize features and image intensities based on imaging modality. Besides promoting disentanglement, the architecture fosters uncooperative learning, offering ability to maintain anatomical structure in a cross-modality reconstruction. Introduction of a modality-invariant structural consistency constraint further enforces faithful embedding of anatomy. To improve training stability and fidelity of synthesized modalities, the ensemble is trained in a relativistic generative adversarial framework incorporating multiscale discriminators. Analyses of priors and network architectures as well as performance validation were performed on computed tomography (CT) and MR pelvis datasets. The proposed method demonstrated robustness against intensity inhomogeneity, improved tissue-class differentiation, and offered synthetic CT in Hounsfield units with intensities consistent and smooth across slices compared to the state-of-the-art approaches, offering median normalized mutual information of 1.28, normalized cross correlation of 0.97, and gradient cross correlation of 0.59 over 324 images.

A Fuzzy Consensus Clustering Algorithm for MRI Brain Tissue Segmentation

Brain tissue segmentation is an important component of the clinical diagnosis of brain diseases using multi-modal magnetic resonance imaging (MR). Brain tissue segmentation has been developed by many unsupervised methods in the literature. The most commonly used unsupervised methods are K-Means, Expectation-Maximization, and Fuzzy Clustering. Fuzzy clustering methods offer considerable benefits compared with the aforementioned methods as they are capable of handling brain images that are complex, largely uncertain, and imprecise. However, this approach suffers from the intrinsic noise and intensity inhomogeneity (IIH) in the data resulting from the acquisition process. To resolve these issues, we propose a fuzzy consensus clustering algorithm that defines a membership function resulting from a voting schema to cluster the pixels. In particular, we first pre-process the MRI data and employ several segmentation techniques based on traditional fuzzy sets and intuitionistic sets. Then, we adopted a voting schema to fuse the results of the applied clustering methods. Finally, to evaluate the proposed method, we used the well-known performance measures (boundary measure, overlap measure, and volume measure) on two publicly available datasets (OASIS and IBSR18). The experimental results show the superior performance of the proposed method in comparison with the recent state of the art. The performance of the proposed method is also presented using a real-world Autism Spectrum Disorder Detection problem with better accuracy compared to other existing methods.

Volume and surface coil simultaneous reception (VSSR) method for intensity inhomogeneity correction in MRI.

Addressing intensity inhomogeneity is critical in magnetic resonance imaging (MRI) because associated errors can adversely affect post-processing and quantitative analysis of images (i.e., segmentation, registration, etc.), as well as the accuracy of clinical diagnosis. Although several prior methods have been proposed to eliminate or correct intensity inhomogeneity, some significant disadvantages have remained, including alteration of tissue contrast, poor reliability and robustness of algorithms, and prolonged acquisition time. In this study, we propose an intensity inhomogeneity correction method based on volume and surface coils simultaneous reception (VSSR). The VSSR method comprises of two major steps: 1) simultaneous image acquisition from both volume and surface coils and 2) denoising of volume coil images and polynomial surface fitting of bias field. Extensive in vivo experiments were performed considering various anatomical structures, acquisition sequences, imaging resolutions, and orientations. In terms of correction performance, the proposed VSSR method was comparatively evaluated against several popular methods, including multiplicative intrinsic component optimization and improved nonparametric nonuniform intensity normalization bias correction methods. Experimental results show that VSSR is more robust and reliable and does not require prolonged acquisition time with the volume coil. The VSSR may be considered suitable for general implementation.