Fuzzy Region Research Articles

Multi-level uncertainty aware learning for semi-supervised dental panoramic caries segmentation

Dental caries segmentation based on oral panoramic medical images is a demanding medical task. However, it is challenging to determine imaging diagnosis results as the actual pathological changes under certain conditions, especially in the suspected pathological locations that distinguish the artifacts from the lesions. This paper proposes introducing the semi-supervised learning framework to redistribute the collected cases to set up exact lesion areas and divide suspected areas with uncertainties into unlabeled data instead of reducing dataset labeling costs. Consistent regularization learning allows us to exploit the feature of ambiguous lesions to optimize decision boundaries. Unfortunately, the labels generated by the general SSL method for fuzzy regions are volatile, especially encountering no specific morphological rules and various scales that fluctuate more seriously, making a devastating impact on the regularization training relying on steady prediction. For such a challenge, we consider places that always form stable predictions under various disorders are often highly deterministic. Therefore, we propose introducing multi-level disturbances, including noise, iterative, and multi-scale disturbances. We simultaneously present in-depth supervision training on the multi-layer decoder to form stable and consistent multi-scale intermediate predictions, which significantly enrich the number of samples gathered in the following integration process. Then, Monte Carlo sampling is applied to integrate multi-level prediction results to assemble a more robust uncertainty mask map, which gradually clarifies the inter-class feature representation of actual lesions and artifacts. The extended experiment shows that we have successfully and effectively improved the performance of the caries segmentation task by using such uncertain lesion sample features. The dataset is now public, and the code is available at https://github.com/Zzz512/MLUA.

BRAIN TUMOR DETECTION AND BRAIN TUMOR AREA CALCULATION WITH MATLAB

Brain tumors that impair the functionality of the person in daily life occur for many different reasons. Treatment of a brain tumor depends on accurately identifying the type, location, size and boundaries of the tumor. Magnetic Resonance Imaging (MRI) technique is used to diagnose the disease. However, this method cannot detect tumors below a certain size due to its nature. The aim of this study is to calculate the area of the tumor region through the successful method after determining which of the Fuzzy C-Means (FCM), Herbaceous Method, Region Growing and Self-Organizing Maps (SOM) methods are more successful in the analysis of MR images. The threshold values of the algorithms used, the number of clusters and the similarity coefficients of jaccard and dice were determined one by one by changing the index codes in the software.The highest similarity index was found in the K-means 10 cluster numbered segmentation in all trials.In general, K-means and Very Grassy Threshold gave very close results. In this context, advanced imaging technique was used by separating the MR image; Tumor spots and brain fluids were detected. Fuzzy C Mean (FCM) was found to be the best method during detection. Brain fluid pushes segmentations used in area calculations to miscalculate. For this reason, while calculating the tumor area, the brain fluids that appear in white spots are completed by point filling. Then, after the tumor zone was identified, the area of this region was used to produce the volume of the region by using Watershed, Graph-Cut and Active Counter segments. It is aimed to determine the number of tumors in which the tumor is in the detection area.

Attribute reduction for hybrid data based on fuzzy rough iterative computation model

A hybrid information system (HIS) means an information system (IS) with categorical attributes, real-valued attributes and missing values. It is more difficult to deal with an HIS than an IS with a type of attributes. This paper studies attribute reduction in an HIS by means of fuzzy rough iterative computation model (FRIC-model). To more accurately express the difference between objects in an HIS, a novel distance function on the object set is first constructed. Then, fuzzy symmetric relations on the object set based on this function are established and fuzzy rough approximations are defined. Next, some evaluation functions that represent the classification ability of attribute subsets, such as fuzzy positive regions, dependency functions and attribute importance functions are introduced. After then, FRIC-model is given and an attribute reduction algorithm for hybrid data based on this model is designed. At last, numerical experiments are carried out to estimate the performance of the designed algorithm. The experimental results show that the designed algorithm is more effective than some existing algorithms.

Feature selection for interval-valued data via FRIC-model

Feature selection is one basic technology for data mining. This paper investigates feature selection for interval-valued data via fuzzy rough iterative computation model (FRIC-model). To depict the similarity between samples in an interval-valued decision information system (IVDIS), the fuzzy symmetry relation in an IVDIS is first introduced from the perspective of “The similarity between information values is fed back to the feature set”. After that, several attribute evaluation functions, such as fuzzy positive regions, dependency functions and attribute importance functions are defined. Subsequently, FRIC-model for interval-valued data is established by using the iterations of these functions. Next, An feature selection algorithm in an IVDIS based on this model is presented. Lastly, numerical experiments and statistics tests are carried out to estimate the performance of the presented algorithm. The experimental results illustrate that the presented algorithm maintains high classification accuracy, and does not occupy too much memory. These findings will provide new perspective for feature selection in an IVDIS.

Output Feedback Control of TS Fuzzy Systems with Unmeasured Nonlinearities via a Local Separation Principle

This paper presents a fuzzy observer-based control framework for TS fuzzy models with nonlinear consequents (N-TS). The N-TS fuzzy model is constructed such that only measured nonlinearities are considered to obtain the fuzzy membership functions, and all unmeasured nonlinearities are modeled as local consequent parts. Based on the differential mean value theorem, the closed-loop system is rewritten in an adequate format, which allows proving the separation principle for asymptotic TS fuzzy observer-based control design. Sufficient design conditions, expressed in terms of linear matrix inequalities, are derived via delayed nonquadratic Lyapunov functions for the synthesis of both stabilizing fuzzy controller and Luenberger-like fuzzy observer. Using Lyapunov-based arguments of a local separation principle, a constructive procedure is provided to maximize the closed-loop domain of attraction (DoA) estimation, which is guaranteed to be included inside the predefined fuzzy modeling region. Numerical examples are given to illustrate the effectiveness of the proposed approach.

WITHDRAWN: FUZZY BASED REGION GROWING SEGMENTATION WITH MODIFIED LEAST SQUARE FITTING ALGORITHM FOR AUTOMATIC FETAL GROWTH ESTIMATION

During pregnancy monitoring, the fetal ultrasound image can provide valuable diagnostic information. Due to low soft tissue contrast, speckle, and shadowing effects, manual segmentation of subcortical structures in ultrasonic volumes is time-consuming and difficult. For this reason, we developed Discrete Wavelet Transform (DWT) based fuzzy region growing segmentation scheme. Considering the aforementioned facts, the present study deals with automatic estimation of gestation age using the head circumference data obtained/ retrieved from the ultrasonic images of the fetus. The proposed work utilizes the DWT based anisotropic diffusion filtering method to enhance the de-noised image resolution. Further implementation of genetic algorithm based Least Square Fitting (LSF) algorithm to find bi-parietal distance (BPD) and head circumference (HC) which will enable the calculate of gestation age. The fetal age is automatically estimated using K-Nearest Neighbour (KNN) Classifier from the standard reference values. By deploying fuzzy based region growing segmentation technique this study attempts to aid unique and effective segmentation process. Apart from highlighting the uniqueness of this method, from other existing techniques, this study elaborately discusses the steps and mechanisms involved in various stages of image processes starting from effective denoising to fetal age determination.

Improved Salp Swarm Optimization-based Fuzzy Centroid Region Growing for Liver Tumor Segmentation and Deep Learning Oriented Classification

Due to heterogenous shape of liver, the segmentation and classification of liver is challenging task. Therefore, Computer-Aided Diagnosis (CAD) is employed for predictive decision making for liver diagnosis. The major intuition of this paper is to detect liver cancer in a precise manner by automatic approach. The developed model initially collects the standard benchmark LiTS dataset, and image preprocessing is done by three techniques like Histogram equalization for contrast enhancement, and median filtering and Anisotropic diffusion filtering for noise removal. Further, the Adaptive thresholding is adopted to perform the liver segmentation. As a novelty, optimized Fuzzy centroid-based region growing model is proposed for tumor segmentation in liver. The main objective of thistumor segmentation model is to maximize the entropy by optimizing the fuzzy centroid and threshold of region growing using Mean Fitness-based Salp Swarm Optimization Algorithm (MF-SSA). From segmented tumor, the features like Local Directional Pattern (LDP) and Gray Level Co-occurrence Matrix (GLCM) are extracted. The extracted features are given as input to NN, and segmented tumor is given to Convolutional Neural Network (CNN). The AND bit operation to both of the outputs obtained from NN and CNN confirms the healthy and unhealthy CT images. Since the number of hidden neurons makes an effect on final classification output, the optimization of neurons is done using MF-SSA. From the experimental analysis, it is confirmed that the proposed model is better as compared with state of art results of previous study can assist radiologists in tumor diagnosis from CT scan images.

Feature extraction for margin lines using region growing with a dynamic weight function in a one-point bidirectional path search

Abstract A margin line, defined as the boundary curve of the contact area between a prepared tooth and a dental restoration, considerably influences the end shape of the dental restoration design. Most studies that have extracted margin lines from mesh models representing prepared teeth have faced convergence problems in the path search and therefore pose the inconvenience of specifying multiple input points as intermediate goal points. To overcome these problems, in this study, we propose a bidirectional path-search algorithm using a single input point. First, the algorithm classifies all nodes in a mesh model into a margin or fuzzy region to increase search efficiency. Then, the search starts from one point and proceeds on two paths in opposite directions, using the current node of the opposite path as the temporary goal of the currently searched path. During the search, a dynamic evaluation function that varies weights according to the region type is employed to improve the path convergence. Finally, to increase the practicality of the algorithm, the jagged initial margin line is converted into a smooth spline curve using an energy-minimization function specialized for margin lines. To evaluate the proposed algorithm, margin lines extracted from various types of prepared teeth are demonstrated and compared with those created using some relevant previous works and a commercial dental computer-aided design (CAD) system. The comparison verified that accurate margin lines could be calculated with only one input point using the proposed algorithm. Moreover, the proposed algorithm showed better performance for crown and inlay/only experimental models compared with a commercial dental CAD system under the same conditions.

Hierarchical Fuzzy Rough Approximations With Three-Way Multigranularity Learning

The approximation learning of fuzzy concepts associated with fuzzy rough sets and three-way decisions is a useful technology for the representation, learning, and transformation of uncertain knowledge. However, how to integrate the advantages of fuzzy rough approximations and three-way approximations has rarely been carefully investigated so far. In this article, we focus on exploring the connection and interplay of these two methods, and further propose the hierarchical fuzzy rough approximations in the dynamic fuzzy open-world environment. We utilize the temporal-spatial perspectives of three-way decisions to construct multigranularity structures and implement multigranularity learning. The temporality of data and the spatiality of model parameters are both considered in such frameworks. Subsequently, we discuss the interpretation and representation of fuzzy three-way regions in fuzzy rough sets with some definitions and properties. The max, double, and aggregated evaluation-based models of three-way approximations are proposed to gradually transform a fuzzy concept to a crisp concept by the time-evolving attributes. Finally, the comparative experimental results between static and dynamic approaches demonstrate the effectiveness of our proposed hierarchical approximation learning models.

Locating dust and molecules in the inner circumstellar environment of R Sculptoris with MATISSE

Context. Asymptotic giant branch (AGB) stars are one of the main sources of dust production in the Galaxy. However, it is not yet clear what this process looks like and where the dust happens to be condensing in the circumstellar environment. Aims. By characterizing the location of the dust and the molecules in the close environment of an AGB star, we aim to achieve a better understanding the history of the dust formation process. Methods. We observed the carbon star R Scl with the thermal-infrared VLTI/MATISSE instrument in L- and N-bands. The high angular resolution of the VLTI observations (as small as 4.4 mas in the L-band and 15 mas in the N-band with ATs), combined with a large uv-plane coverage allowed us to use image reconstruction methods. To constrain the dust and molecules’ location, we used two different methods: one using MIRA image reconstruction algorithm and the second using the 1D code RHAPSODY. Results. We found evidence of C2H2 and HCN molecules between 1 and 3.4 R* which is much closer to the star than the location of the dust (between 3.8 and 17.0 R*). We also estimated a mass-loss rate of 1.2 ± 0.4 × 10−6 M⊙ yr−1. In the meantime, we confirmed the previously published characteristics of a thin dust shell, composed of amorphous carbon (amC) and silicon carbide (SiC). However, no clear SiC feature has been detected in the MATISSE visibilities. This might be caused by molecular absorption that can affect the shape of the SiC band at 11.3 µm. Conclusions. The appearance of the molecular shells is in good agreement with predictions from dynamical atmosphere models. For the first time, we co-located dust and molecules in the environment of an AGB star. We confirm that the molecules are located closer to the star than the dust. The MIRA images unveil the presence of a clumpy environment in the fuzzy emission region beyond 4.0 R*. Furthermore, with the available dynamic range and angular resolution, we did not detect the presence of a binary companion. To solve this problem, additional observations combining MATISSE and SAM-VISIR instrument should enable this detection in future studies.

GFNet: Automatic segmentation of COVID-19 lung infection regions using CT images based on boundary features

In early 2020, the global spread of the COVID-19 has presented the world with a serious health crisis. Due to the large number of infected patients, automatic segmentation of lung infections using computed tomography (CT) images has great potential to enhance traditional medical strategies. However, the segmentation of infected regions in CT slices still faces many challenges. Specially, the most core problem is the high variability of infection characteristics and the low contrast between the infected and the normal regions. This problem leads to fuzzy regions in lung CT segmentation. To address this problem, we have designed a novel global feature network(GFNet) for COVID-19 lung infections: VGG16 as backbone, we design a Edge-guidance module(Eg) that fuses the features of each layer. First, features are extracted by reverse attention module and Eg is combined with it. This series of steps enables each layer to fully extract boundary details that are difficult to be noticed by previous models, thus solving the fuzzy problem of infected regions. The multi-layer output features are fused into the final output to finally achieve automatic and accurate segmentation of infected areas. We compared the traditional medical segmentation networks, UNet, UNet++, the latest model Inf-Net, and methods of few shot learning field. Experiments show that our model is superior to the above models in Dice, Sensitivity, Specificity and other evaluation metrics, and our segmentation results are clear and accurate from the visual effect, which proves the effectiveness of GFNet. In addition, we verify the generalization ability of GFNet on another “never seen” dataset, and the results prove that our model still has better generalization ability than the above model. Our code has been shared at https://github.com/zengzhenhuan/GFNet.

Retraction Note to: Rough fuzzy region based bounded support fuzzy C-means clustering for brain MR image segmentation

Retraction Note to: Rough fuzzy region based bounded support fuzzy C-means clustering for brain MR image segmentation

Automatic lung segmentation in chest X-ray images using improved U-Net

The automatic segmentation of the lung region for chest X-ray (CXR) can help doctors diagnose many lung diseases. However, extreme lung shape changes and fuzzy lung regions caused by serious lung diseases may incorrectly make the automatic lung segmentation model. We improved the U-Net network by using the pre-training Efficientnet-b4 as the encoder and the Residual block and the LeakyReLU activation function in the decoder. The network can extract Lung field features efficiently and avoid the gradient instability caused by the multiplication effect in gradient backpropagation. Compared with the traditional U-Net model, our method improves about 2.5% dice coefficient and 6% Jaccard Index for the two benchmark lung segmentation datasets. Our model improves about 5% dice coefficient and 9% Jaccard Index for the private lung segmentation datasets compared with the traditional U-Net model. Comparative experiments show that our method can improve the accuracy of lung segmentation of CXR images and it has a lower standard deviation and good robustness.

The Influence of the Elemental Composition, Crystal Structure, and Grain Structure of the Ferro-Piezoceramics of Various Degrees of the Ferro-Hardness on the Stability of the Polarized State

Ferro-piezoceramic materials (FPCM) with different degrees of ferrohardness were fabricated by double solid-phase synthesis followed by the sintering technique using hot pressing method. The X-ray studies carried out in a wide temperature range showed that with increasing temperature, each of the studied FPCM undergoes a series of phase transformations, accompanied by a change in the symmetry of the unit cell. In this case, near the phase transition to the nonpolar cubic phase, in each of the FPCM, the formation of a fuzzy symmetry region is observed, which is characterized by weak distortions and temperature–time instability of the crystal structure. The study of the piezoelectric modulus d33 in the quasi-static regime as a function of temperature made it possible to reveal the different nature of its behavior in materials of various degrees of ferrohardness. It was shown that the conservation of the state in ferrosoft materials above the Curie temperature is associated with the relaxation nature of the change in their properties, the existence of a region of fuzzy symmetry (noncubic phase) in them above the Curie temperature, and increased inertia of the system. The expediency of taking into account the presented results in the development of electromechanical converters based on FPCM of various degrees of ferrohardness, operated under temperature effects, including cyclic ones, was shown.

A Novel Brain Tumor Segmentation Method Based on Im-proved Spatial Attention Mechanism and Multi-path Deep Neural Network

Preoperative Magnetic Resonance Image (MRI) brain tumor diagnosis is an effective technical approach. To accurately segment tumor regions, we propose a novel brain tumor segmentation method based on improved Spatial Attention mechanism and Multi-path neural network (SAMPU-Net). Firstly, we propose a multipath input method to extract feature information of different scales by using convolution kernels of different sizes, so as to fully extract MRI feature information. Secondly, we improve the spatial attention mechanism by adding convolution layer of pyramid structure to it to obtain the features of different receptive fields. In the convolution layer of this pyramid structure, the larger the convolution kernel is, the more global features will be extracted; conversely, the smaller the convolution kernel is, the more local features will be extracted. Thirdly, we use more multi-mode MRI information to segment the brain tumor images. In practical application, due to the fuzzy tumor regions in some MRIs, we use the method of restricted contrast adaptive histogram equalization to perform local enhancement of images. The proposed model and several other mainstream segmentation methods are trained and tested on the BraTS2019 public dataset. Experimental results indicate that using our method, the Dice coefficient of tumor core and tumor enhancement region is increased by 2.4% and 1.3% respectively, and our proposed method has better segmentation effect than other methods.

Dynamic maintenance of variable precision fuzzy neighborhood three-way regions in interval-valued fuzzy decision system

Dynamic maintenance of variable precision fuzzy neighborhood three-way regions in interval-valued fuzzy decision system

Satellite coverage traffic volume prediction using a new surrogate model

The space-based Automatic Dependent Surveillance-Broadcast (ADS-B) can significantly improve aircraft traffic efficiency and safety but suffers from the issue of signal collision. A potential solution for this limitation is employing an adaptive ADS-B antenna, e.g., the phased array antenna, which can adjust its coverage area in real-time during the satellite movement according to the satellite coverage traffic volume (SCTV). This paper proposes a new surrogate modeling technique, i.e., subregion radial basis function (RBF), and applies the proposed surrogate to predict the SCTV so that the adaptive strategy for the ADS-B antenna can be designed. Instead of training one model using all training samples as the classical RBF does, the proposed subregion RBF approach first partitions the input space into multiple fuzzy regions, then trains a RBF model in each region, and finally integrates the multiple subregion RBF models in a weighted average manner. By using this partition of unity strategy, the proposed method can greatly reduce the computational complexity of training models meanwhile maintain the interpolation property and high accuracy of RBF models. Further, the proposed subregion RBF is applied to construct a surrogate for the SCTV model to guide the design of the adaptive strategy for the space-based ADS-B. Simulation results show that compared with the classical and state-of-the-art methods, the proposed method has arrived a better balance between predictive accuracy and computational cost. In the simulation scenarios, it is also observed that by using the adaptive strategy, the detection probability of ADS-B receiver can always be greater than 95% and nearly 1/3 of power consumption can be saved. The simulations demonstrate the effectiveness of the proposed method.

Strain rate effects on mechanical behavior and microstructure evolution with the sequential strains of TWIP steel

Dynamic and quasi-static mechanical behaviors and microstructural mechanisms of the TWIP steel were investigated at strain rates from 0.001 to 3000 s-1 by using a split Hopkinson tensile bar paired with interrupted strain fixtures and electron backscattered diffraction technology. The positive strain rate sensitivity of dynamic yield stress is significantly different from that of quasi-static stress at a transition strain rate of about 100 s-1 for the different yield mechanisms. Owing to the high stress from dynamic loading, the smaller twin onset strain and the higher twin fraction as well as the more intense twin boundary-dislocation interaction are responsible for the enhanced strain hardening behavior in the early stage of strain. Nevertheless, the rapid decrease of the strain hardening rate in the later stage and the resulting negative strain hardening rate sensitivity are due to the decreasing growth rate of the twin boundary fraction. It is closely associated with the softening effect of adiabatic temperature rise (∼117 °C) by raising the stacking fault energy and inhibiting twinning activity. In addition, the fuzzy regions caused by high-density dislocations within interfaces between twin and grain boundaries may lead to the lower elongation of the dynamic tensile specimen. The grains with a high Schmid factor are conducive to slip rather than twinning, which enables the twin region orientation for slip-mediated plastic deformation. This study thus advances the understanding of the strain and strain rate dependence of the mechanical behaviors and twinning mechanisms of TWIP steels for the safety performance of automobiles.

Fuzzy controller synthesis algorithm with complete data on the object state

The application of the classical control theory to nonlinear and linear objects control problems in some cases may not give effective results. Due to this, fuzzy controller synthesis methods for objects controlling have been developed based on fuzzy sets and fuzzy logic methods. A suggested method makes it possible to construct a fuzzy system without the help from human experts. The algorithm for this method consists of three steps: on step 1 the known input data ranges are determined and divided into fuzzy regions to form membership functions; on step 2 the data pairs are formed and then used to generate a fuzzy rules table; on step 3 the membership functions are scaled for specific values, so the fuzzy system works more correct. The method has shown it’s good performance in the underwater vehicle stabilization problem with complete data on the object state.

A fuzzy approach for determining the cognitive spatial location of an object in geographical information system

The work is devoted to the problem of interpretation of fuzzy semantics of cognitive descriptions of spatial relations in natural language and their visualization in a geographic information system (GIS). The solution to the problem of determining the fuzzy spatial location of an object based on vague descriptions of the observer in natural language is considered. The task is relevant in critical situations when there is no way to report the exact coordinates of the observed object, except by describing its location relative to the observer itself. Such a situation may be the result of a crime, terrorist act or natural disaster. An observer who finds itself at the scene transmits a text message, which is a description of the location of the object or place (for example, the crime scene, the location of dangerous objects, the crash site). The semantics of the spatial location of the object can be further extracted from the text message. The proposed fuzzy approach is based on the formalization of the observer's phrases, with which it can describe spatial relations, in the form of a set of linguistic variables that determine the direction and distance to the object. Examples of membership functions for linguistic variables are given. The spatial knowledge base is built on the basis of the phrases of observers and their corresponding fuzzy regions. Algorithms for constructing cognitive regions in GIS have been developed. Methods of their superposition to obtain the final fuzzy location of the object are proposed. An example of the implementation of a fuzzy model for identifying cognitive regions based on vague descriptions of several observers, performed using developed Python scripts integrated into ArcGIS 10.5, is considered.