Aboul Ella Hassanien (Abo)

This article introduces a hybrid approach that combines the advantages of fuzzy sets, pulse coupled neural networks (PCNNs), and support vector machine, in conjunction with wavelet-based feature extraction. An application of breast cancer MRI imaging has been chosen and hybridization approach has been applied to see their ability and accuracy to classify the breast cancer images into two outcomes: normal or non-normal. The introduced approach starts with an algorithm based on type-II fuzzy sets to enhance the contrast of the input images. This is followed by performing PCNN-based segmentation algorithm in order to identify the region of interest and to detect the boundary of the breast pattern. Then, wavelet-based features are extracted and normalized. Finally, a support vector machine classifier was employed to evaluate the ability of the lesion descriptors for discrimination of different regions of interest to determine whether they represent cancer or not. To evaluate the performance of presented approach, we present tests on different breast MRI images. The experimental results obtained, show that the overall accuracy offered by the employed machine learning techniques is high compared with other machine learning techniques including decision trees, rough sets, neural networks, and fuzzy artmap.

Abstract—Extraction of relevant features is of significant importance for brain tumor segmentation systems. In this paper, with the objective of improving brain tumor segmentation accuracy, we present an improved feature extraction component to take advantage of the correlation between intracranial structure deformation and the compression from brain tumor growth. Using 3-dimensional non-rigid registration and deformation modeling techniques, the component is capable of measuring lateral ventricular (LaV) deformation in the volumetric magnetic resonance (MR) images. By verifying the location of the extracted LaV deformation feature data and applying the features on brain tumor segmentation with widely used classification algorithms, the proposed component is evaluated qualitatively and quantitatively with promising results on 11 datasets comprising real patient and simulated images.

Extraction of relevant features is of significant importance for brain tumor segmentation systems. To improve brain tumor segmentation accuracy, the authors present an improved feature extraction component that takes advantage of the correlation between intracranial structure deformation and the compression resulting from brain tumor growth. Using 3D nonrigid registration and deformation modeling techniques, the component measures lateral ventricular (LaV) deformation in volumetric magnetic resonance images. By verifying the location of the extracted LaV deformation feature data and applying the features on brain tumor segmentation with widely used classification algorithms, the authors evaluate the proposed component qualitatively and quantitatively with promising results on 11 datasets comprising real and simulated patient images.

Actors in social networks tend to form community groups based on common location, interests, occupation, etc. Communities play special roles in the structure–function relationship; therefore, detecting such communities can be a way to describe and analyze such networks. However, the size of those networks has grown tremendously with the increase of computational power and data storage. While various methods have been developed to extract community structures, their computational cost or the difficulty to parallelize existing algorithms make partitioning real networks into communities a challenging problem. In this paper, we introduce a generative process to model the interactions between social network’s actors. Through unsupervised learning using expectation maximization, we derive an efficient and fast community detection algorithm based on Bayesian network and expectation maximization (BNEM). We show that BNEM algorithm can infer communities within directed or undirected networks, and within weighted or un-weighted networks. We also show that the algorithm is easy to parallelize. We then explore and analyze the result of the BNEM method. Finally, we conduct a comparative analysis with other well-known methods in the fields of community detection.

In this paper, we propose a fragile watermarking hybrid approach using rough set kmeans and exponential particle swarm optimization (EPSO) systems. It is based on a block-wise dependency mechanism which can detect any alterations made to the protected image. Initially, the input image is divided into blocks with equal size in order to improve image tamper localization precision. Then feature sequence is generated by applying rough k-means and EPSO clustering to create the relationship between all image blocks and cluster
all of them since EPSO is used to optimize the parameters of rough k-means. Both feature sequence and generated secret key are used to construct the authentication data. Each resultant 8-bit authentication data is embedded into the eight least significant bits (LSBs) of the corresponding image block. We gives experimental results which show the feasibility of using these optimization algorithms for the fragile watermarking and demonstrate the
accuracy of the proposed approach. The performance comparison of the approach was also realized. The performance of a fragile watermarking approach has been improved in this paper by using exponential particle swarm optimization (EPSO) to optimize the rough kmean parameters. The proposed approach can embed watermark without causing noticeable visual artifacts, and does not only achieve superior tamper detection in images accurately,
it also recovers tampered regions effectively. In addition, the results show that the proposed approach can effectively thwart different attacks, such as the cut-and paste attack and collage attack, while sustaining superior tamper detection and localization accuracy.