Mohamed Abu ElSoud

Neutrosophic Sets and Fuzzy C-Means Clustering for Improving CT Liver Image Segmentation

In this paper, an improved segmentation approach based on Neutrosophic sets (NS) and fuzzy c-mean clustering (FCM) is proposed. An application of abdominal CT imaging has been chosen and segmentation approach has been applied to see their ability and accuracy to segment abdominal CT images. The abdominal CT image is transformed into NS domain, which is described using three subsets namely; the percentage of truth in a subset T, the percentage of indeterminacy in a subset I, and the percentage of falsity in a subset F. The entropy in NS is defined and employed to evaluate the indeterminacy. Threshold for NS image is adapted using Fuzzy C-mean algorithm. Finally, abdominal CT image is segmented and liver parenchyma is selected using connected component algorithm. The proposed approach denoted as NSFCM and compared with FCM using Jaccard Index and Dice Coefficient. The experimental results demonstrate that the proposed approach is less sensitive to noise and performs better on nonuniform CT images.

Automatic liver parenchyma segmentation system from abdominal CT scans using hybrid techniques

In this paper, a multi–layer heuristic approach is introduced to segment liver region from other tissues in multi–slice CT images. Image noise is a principal factor which hampers the visual quality of medical images and can therefore lead to misdiagnosis. To address this issue, we first utilise an algorithm based on median filter to remove noise and enhance the contrast of the CT image. This is followed by performing an adaptive threshold algorithm and morphological operators to preserve the liver structure and remove the fragments of other organs. Then, connected component labelling algorithm was applied to remove false positive regions and focused on liver region. To evaluate the performance of the proposed system, we present tests on different liver CT scans images. The experimental results show that the overall accuracy offered by the employed system is high compared with other related works as well as very fast which segment liver from abdominal CT in less than 0.6 s/slice.

Automatic mammogram segmentation and computer aided diagnoses for breast tissue density according to BIRADS dictionary

It is widely accepted in the medical community that breast tissue density is an important risk factor for the development of breast cancer. Thus, the development of reliable automatic methods for classification of breast tissue is justified and necessary. Recent studies have shown that their sensitivity is significantly decreased as the density of the breast is increased. The internal density of the breast is a parameter that clearly affects the performance of segmentation algorithms in defining abnormal regions. In this study, the breast region was extracted from background, and pectoral muscle was suppressed. We review different methods for computing tissue density parameter. An attempt was made to combine the individual scores from different techniques in order to compensate their individual weaknesses and to preserve their strength. We also present and exhaustively evaluate algorithms using computer vision techniques. We obtained accuracy as high as 94% of automatic agreement classification.

Feature Selection Approach Based on Social Spider Algorithm: Case Study on Abdominal CT Liver Tumor

This paper addresses a new subset feature selection performed by new Social Spider Optimization algorithm (SSOA) to find optimal regions of the complex search space through the interaction of individuals in the population. SSOA is a new evolutionary computation technique which mimics the behavior of cooperative social-spiders based on the biological laws of the cooperative colony. The performance of SSOA associated with two reasons: (a) operators allow to increasing find the global optima in the search space, and (b) division of the population into male and female, provides the use of different rates between exploration and exploitation during the evolution process. A theoretical analysis on abdominal CT liver tumor dataset that models the number of correctly classified data is proposed using Confusion Matrix, Precision, Recall, and accuracy. The results show that the mechanism of SSOA provides very good exploration, local minima avoidance, and exploitation simultaneously.

An intelligent approach for galaxies images classification

This article presents an intelligent automatic approach for galaxies images classification based on Artificial Neural Network (ANN) and moment-based features extraction algorithms. The proposed approach consists of three phases; namely, image denoising, feature extraction, and classification phases. For the denoising phase, noise pixels are removed from input images, then input galaxy image is normalized to a uniform scale and Hu seven invariant moment algorithm is applied to reduce the dimensionality of the feature space during the feature extraction phase. Finally, during the classification phase, Self-Organize Feature Maps (SOFMs) and Time Lag Recurrent Networks (TLRNs) algorithms are utilized for classifying the input galaxies images into one of four obtained source catalogue types. Experimental results showed that SOFMs provided better classification results than having TLRNs applied. It is also concluded that a small set of features is sufficient to classify galaxy images and provide a fast classification.

Automated classification of galaxies using invariant moments

Classification and identification of galaxy shape is an important issue for astronauts since it provides valuable information about the origin and the evolution of the universe. Statistical invariant features that are functions of moments have been used as global features of galaxy images in their pattern recognition. In this paper, an automated training based recognition system that can compute the statistical invariant features for different galaxy shapes is investigated. The proposed algorithm is robust, regardless of orientation, size and position of the galaxy inside the image. Feature vectors are computed via nonlinear moment invariant functions for each galaxy shape. After feature extraction, the recognition performance of classifier in conjunction with these moment---based features is introduced. Computer simulations show that Galaxy images are classified with an accuracy of about 90% compared to the human visual classification system.

Computational Intelligence Optimization Algorithm Based on Meta-heuristic Social-Spider: Case Study on CT Liver Tumor Diagnosis

Feature selection is an importance step in classification phase and directly affects the classification performance. Feature selection algorithm explores the data to eliminate noisy, redundant, irrelevant data, and optimize the classification performance. This paper addresses a new subset feature selection performed by a new Social Spider Optimizer algorithm (SSOA) to find optimal regions of the complex search space through the interaction of individuals in the population. SSOA is a new natural meta-heuristic computation algorithm which mimics the behavior of cooperative social-spiders based on the biological laws of the cooperative colony. Different combinatorial set of feature extraction is obtained from different methods in order to keep and achieve optimal accuracy. Normalization function is applied to smooth features between [0,1] and decrease gap between features. SSOA based on feature selection and reduction compared with other methods over CT liver tumor dataset, the proposed approach proves better performance in both feature size reduction and classification accuracy. Improvements are observed consistently among 4 classification methods. A theoretical analysis that models the number of correctly classified data is proposed using Confusion Matrix, Precision, Recall, and Accuracy. The achieved accuracy is 99.27%, precision is 99.37%, and recall is 99.19%. The results show that, the mechanism of SSOA provides very good exploration, exploitation and local minima avoidance.

An Improved Prediction Approach for Progression of Ocular Hypertension to Primary Open Angle Glaucoma

In this paper, we present an improved prediction model for progression of ocular hypertension to primary open angle glaucoma using a random forest classification approach. Our model comprises two phases: risk factor calculation and prediction. We start by calculating the risk factors associated with the outcome, followed by a prediction phase that utilises a random forest approach for classification into one of four obtained classes, namely low, mid, high, and moderate. Experimental results show that the employed random forest classifier provides better prediction accuracy compared to other machine learning techniques including Bayes net, multi-layer perceptron, radial basis function and naive Bayes tree classifiers.

Automatic liver Parenchyma segmentation from abdominal CT images

This article introduces hybrid automatic liver Parenchyma segmentation approach from abdominal CT images. The proposed approach consist of four main phases. Firstly, preprocessing phase which converts CT image into binary image using adaptive threshold method that examine the intensity values of the local neighborhood of each pixel. Then, the second phase is to apply multi-scale morphological operators to filter tissues nearby liver and to preserve the liver structure and remove the fragments of other organs. The third phase is a post-processing that uses connected component labeling algorithm (CCL) to remove small objects and false positive regions. The algorithm is tested using two different datasets and the experimental results obtained, show that the proposed approach are promising which could segment liver from abdominal CT in less than 0.6 s/slice and the overall accuracy obtained by the proposed approach is 93%.

Optimal feature level fusion based ANFIS classifier for brain MRI image classification

The cases identified with Brain tumor have increased with respect to time owing to various reasons. One of the major challenging issues can be defined by incorporating image processing along with data mining models as classification approach. There are various procedures as of now exhibited for segmentation of brain tumor effectively. In any case, it is as yet unequivocal to distinguish the brain tumor from MR images. In this new tumor classifying, considering two significant models, such as Feature Selection (FS) and Machine Learning classification techniques, are extremely valuable for distinguishing and visualizing the tumor in the MRI brain images; it is classified using Adaptive Neuro‐Fuzzy Interface System (ANFIS). For better classification of image, Optimal Feature Level Fusion (OFLF) is considered to fuse low and high‐level feature of brain image; from this analysis, the images are classifying as Benign or Malignant. From this implementation of medical images, the experiment results are evaluating performance metrics are compared existing classifiers. From the proposed MRI image classification process the accuracy as 96.23%, sensitivity as 92.3%, and specificity as 94.52%, compared to existing classifier. It is in the working platform of MATLAB that this proposed methodology is implemented.