Uneven Intensity Distribution Research Articles

Effective hybrid attention network based on pseudo-color enhancement in ultrasound image segmentation

Ultrasound image segmentation plays a vital role in the early diagnosis of human diseases. It helps diagnose many diseases, such as breast cancer, hemangioma, and other gynecological disorders. However, the intrinsic imaging characteristics of ultrasound images result in substantially lower resolution and clarity than CT, MRI, and other imaging modalities, and they are sensitive to interference from external influences. With its inherent artifacts, blurred lesion boundaries, and uneven intensity distribution, ultrasound images present a challenging task when it comes to segmenting lesion areas accurately. In recent years, convolutional neural networks (CNNs) have achieved remarkable results in medical image segmentation tasks. However, CNNs are limited in capturing the remote dependencies of the input image, leading to degraded accuracy in segmenting ultrasound lesions. In this paper, we developed a deep convolutional neural network that incorporates the pseudo-color enhancement algorithm and hybrid attention modules that enhance the network’s ability to extract fine features and remote modeling capabilities. We propose a novel multi-scale channel attention-based decoder that efficiently uses the feature maps from the encoder as a complement and fuses them with the upsampled feature maps. The hybrid attention combination captures cross-channel interactions efficiently and enhances the context modeling capability, further improving the extraction of coarse and delicate features, and resulting in significant performance improvements. We found that the dice performance improved by 2.54%, 2.47%, 1.39%, 0.99%, and 1.23% on the BUL, BUSI, Hemangioma, BP, and VUI. Results from four public datasets and one self-collected dataset indicate that the proposed method outperforms other medical image segmentation methods for ultrasound image lesion segmentation.

An optimization method of rough surface adapted small rectangular LD laser source based on combination of Powell prism and cylindrical lenses

A combination optimization method was proposed to design a small-size rectangular laser spot and decrease the measurement error caused by the uneven intensity distribution of LD laser source. First, the intensity distribution of LD was analyzed and the combination lens structure of Powell prism and cylindrical lenses was proposed. The equations were established to calculate the profile parameters of lenses. And an illuminance loss model δΔED±R was built to determine the values of laser spot size, working reference and range. Secondly, the specific design process of our method was given, the 3D model was established and simulated in ZEMAX to further improve the imaging uniformity. Finally, the collimation, uniformity and location deviation of the spot were discussed. The average uniformity was 95.626%, and the difference of location deviation on shot peened surfaces was 0.004 μm in x direction and 0.121 μm in y direction when compared with the white ceramic surface.

Dilated convolution network with edge fusion block and directional feature maps for cardiac MRI segmentation.

Cardiac magnetic resonance imaging (MRI) segmentation task refers to the accurate segmentation of ventricle and myocardium, which is a prerequisite for evaluating the soundness of cardiac function. With the development of deep learning in medical imaging, more and more heart segmentation methods based on deep learning have been proposed. Due to the fuzzy boundary and uneven intensity distribution of cardiac MRI, some existing methods do not make full use of multi-scale characteristic information and have the problem of ambiguity between classes. In this paper, we propose a dilated convolution network with edge fusion block and directional feature maps for cardiac MRI segmentation. The network uses feature fusion module to preserve boundary information, and adopts the direction field module to obtain the feature maps to improve the original segmentation features. Firstly, multi-scale feature information is obtained and fused through dilated convolutional layers of different scales while downsampling. Secondly, in the decoding stage, the edge fusion block integrates the edge features into the side output of the encoder and concatenates them with the upsampled features. Finally, the concatenated features utilize the direction field to improve the original segmentation features and generate the final result. Our propose method conducts comprehensive comparative experiments on the automated cardiac diagnosis challenge (ACDC) and myocardial pathological segmentation (MyoPS) datasets. The results show that the proposed cardiac MRI segmentation method has better performance compared to other existing methods.

TSASNet: Tooth segmentation on dental panoramic X-ray images by Two-Stage Attention Segmentation Network

Tooth segmentation acts as a crucial and fundamental role in dentistry for doctors to make diagnosis and treatment plans. In this paper, we propose a Two-Stage Attention Segmentation Network (TSASNet) on dental panoramic X-ray images to address the issues suffered in the tooth boundary and tooth root segmentation task which are caused by the low contrast and uneven intensity distribution. We firstly adopt an attention model which is embedded with global and local attention modules to roughly localize the tooth region in the first stage. Without any interactive operator, the attention model so constructed can automatically aggregate pixel-wise contextual information and identify coarse tooth boundaries. To better obtain final boundary information, we use a fully convolutional network as the second stage to further segment the real tooth area from the attention maps obtained from the first stage. The effectiveness of TSASNet is substantiated on the benchmark dataset containing 1,500 dental panoramic X-ray images, our proposed method achieves 96.94% of accuracy, 92.72% of dice and 93.77% of recall, significantly superior to the current state-of-the-art methods.

Correction of errors due to uneven intensity distribution and inaccurate phase shifts in prism-based simultaneous phase-shifting interferometry.

Simultaneous phase-shifting interferometry (SPSI) is a suitable technique for dynamic surface measurement due to its rapid measurement capability. Based on whether there are obvious multiple phase-shifting channels in the setup, it can be divided into the multi-channel type and the pixelated phase-mask type. For the former type, two important error sources are the uneven intensity distribution among different phase-shifting channels and inaccurate phase shifts in each phase-shifting channel. In order to diminish the influence of these two error sources, several methods have been proposed. However, in some of these methods, only one error source of the two was considered and corrected. In the methods where the two error sources were both considered, some assumptions such as the uniform background intensity and fringe contrast are needed, so a complete and universal suppression method for both error sources is still needed in actual applications. In this paper, for a prism-based SPSI, we proposed such an error-correction method that mainly contains the following contributions. First, the intrinsic parameters of the system, including intensity distribution coefficients and relative phase shifts among phase-shifting channels, are calibrated in advance. We also checked the uniformity of phase shifts in each phase-shifting channel. Second, based on the obtained parameters, a corrected four-step phase-shifting algorithm is deduced to recover the more accurate measured phase. Results of both simulation and experiment verify the effectiveness of the proposed method.

Three-dimensional surface imaging by multi-frequency phase shift profilometry with angle and pattern modeling for system calibration

In this paper, we present a 3D surface imaging system based on the well-known phase shift profilometry. To yield the analytical solutions, four shifted phases and three high carrier frequencies are used to compute the phase map and reduce the noises that are caused by the inherent optical aberrations and external influences, e.g. different illumination light sources, uneven intensity distribution and automatic image processing algorithms. To reduce the system noise, we propose to model the pattern of the calibration grid in a virtual space. To obtain the modeled pattern, we use a plane to intercept the rays that are modeled by the proposed angle modeling method. In the world coordinate system, the angle and the pattern are computed based on the calibration data. A registration method is used to transform the modeled pattern in the virtual space to the ideal pattern in the world coordinate system by computing the least squared errors between the true points in the modeled pattern and the measured points in the practical pattern. The modeled (true) points are used for re-calibration of the 3D imaging system. Experimental results showed that the measurement accuracy increases considerably and the MSE is reduced from 0.95 mm to 0.65 mm (32% average error decrease) after replacing the measured points with the true points for calibration.

Framelet-based sparse regularization for uneven intensity correction of remote sensing images in a retinex variational framework

Correcting uneven intensity distribution from a single image has long been a challenging problem with remote sensing image. In this paper, an analysis-based sparse prior is employed in the retinex variational framework for the uneven intensity correction of remote sensing images. This sparse regularization model is used to adjust uneven intensity by regularizing the sparsity of the reflectance component under framelet transform. Furthermore, the alternating minimization algorithm and split Bregman method are adopted to solve the framelet-based sparse regularization model. The experiments, with both simulated images and real-life images, show that the proposed model can effectively correct the uneven intensity distribution.

Detection of Wormhole in Maize based on Kernel RGB Image and Lighting Transformation Method

To effectively extract wormhole areas in kernels of maize, the uneven intensity distribution that was produced by the lighting system or by part of the vision system in the image must be transformed. A methodology was developed to convert non-uniform intensity distribution on objects into a uniform intensity distribution. A basically plane image with the wormhole area having a lower gray level than this plane was obtained by using proposed algorithms. Then, the wormhole areas can be easily extracted by a global threshold value. The experimental results with a 99.0% classification rate based on 100 kernel images showed that the proposed algorithm was simple and effective.

Photofragment image analysis using the Onion-Peeling Algorithm

With the growing popularity of the velocity map imaging technique, a need for the analysis of photoion and photoelectron images arose. Here, a computer program is presented that allows for the analysis of cylindrically symmetric images. It permits the inversion of the projection of the 3D charged particle distribution using the Onion Peeling Algorithm. Further analysis includes the determination of radial and angular distributions, from which velocity distributions and spatial anisotropy parameters are obtained. Identification and quantification of the different photolysis channels is therefore straightforward. In addition, the program features geometry correction, centering, and multi-Gaussian fitting routines, as well as a user-friendly graphical interface and the possibility of generating synthetic images using either the fitted or user-defined parameters. Program summary Title of program: Glass Onion Catalogue identifier: ADRY Program Summary URL: http://cpc.cs.qub.ac.uk/summaries/ADRY Program obtainable from: CPC Program Library, Queen's University of Belfast, N. Ireland Licensing provisions: none Computer: IBM PC Operating system under which the program has been tested: Windows 98, Windows 2000, Windows NT Programming language used: Delphi 4.0 Memory required to execute with typical data: 18 Mwords No. of bits in a word: 32 No. of bytes in distributed program, including test data, etc.: 9 911 434 Distribution format: zip file Keywords: Photofragment image, onion peeling, anisotropy parameters Nature of physical problem: Information about velocity and angular distributions of photofragments is the basis on which the analysis of the photolysis process resides. Reconstructing the three-dimensional distribution from the photofragment image is the first step, further processing involving angular and radial integration of the inverted image to obtain velocity and angular distributions. Provisions have to be made to correct for slight distortions of the image, and to verify the accuracy of the analysis process. Method of solution: The “Onion Peeling” algorithm described by Helm [Rev. Sci. Instrum. 67 (6) (1996)] is used to perform the image reconstruction. Angular integration with a subsequent multi-Gaussian fit supplies information about the velocity distribution of the photofragments, whereas radial integration with subsequent expansion of the angular distributions over Legendre Polynomials gives the spatial anisotropy parameters. Fitting algorithms have been developed to centre the image and to correct for image distortion. Restrictions on the complexity of the problem: The maximum image size (1280×1280) and resolution (16 bit) are restricted by available memory and can be changed in the source code. Initial centre coordinates within 5 pixels may be required for the correction and the centering algorithm to converge. Peaks on the velocity profile separated by less then the peak width may not be deconvolved. In the charged particle image reconstruction, it is assumed that the kinetic energy released in the dissociation process is small compared to the energy acquired in the electric field. For the fitting parameters to be physically meaningful, cylindrical symmetry of the image has to be assumed but the actual inversion algorithm is stable to distortions of such symmetry in experimental images. Typical running time: The analysis procedure can be divided into three parts: inversion, fitting, and geometry correction. The inversion time grows approx. as R 3, where R is the radius of the region of interest: for R=200 pixels it is less than a minute, for R=400 pixels less then 6 min on a 400 MHz IBM personal computer. The time for the velocity fitting procedure to converge depends strongly on the number of peaks in the velocity profile and the convergence criterion. It ranges between less then a second for simple curves and a few minutes for profiles with up to twenty peaks. The time taken for the image correction scales as R 2 and depends on the curve profile. It is on the order of a few minutes for images with R=500 pixels. Unusual features of the program: Our centering and image correction algorithm is based on Fourier analysis of the radial distribution to insure the sharpest velocity profile and is insensitive to an uneven intensity distribution. There exists an angular averaging option to stabilize the inversion algorithm and not to loose the resolution at the same time.

Unusual two-dimensional angular distribution of photoelectrons of kish graphite and 1T-TaS 2

Two-dimensional patterns of angle-resolved photoemission are measured on Kish graphite and 1T-TaS 2 for s-polarized synchrotron radiation. Observed symmetry-broken patterns are theoretically discussed. The effect of the dipole selection rules by the s-polarized light is clarified. It is found for the first time that an interference of photoelectron waves from non-equivalent sites provides uneven intensity distribution in different Brillouin zones. The patterns of 1T-TaS 2 measured at 40 eV are remarkably different from those at 22 eV, which is explained by the resonant photoemission in the Ta 5 p core excitation region breaking the dipole selection rule due to the direct recombination type Auger transition.