Deepak Ranjan Nayak

Brain MR image classification using two-dimensional discrete wavelet transform and AdaBoost with random forests

This paper presents an automated and accurate computer-aided diagnosis (CAD) system for brain magnetic resonance (MR) image classification. The system first utilizes two-dimensional discrete wavelet transform (2D DWT) for extracting features from the images. After feature vector normalization, probabilistic principal component analysis (PPCA) is employed to reduce the dimensionality of the feature vector. The reduced features are applied to the classifier to categorize MR images into normal and abnormal. This scheme uses an AdaBoost algorithm with random forests as its base classifier. Three benchmark MR image datasets, Dataset-66, Dataset-160, and Dataset-255, have been used to validate the proposed system. A 5×5-fold stratified cross validation scheme is used to enhance the generalization capability of the proposed scheme. Simulation results are compared with the existing schemes and it is observed that the proposed scheme outperforms others in all the three datasets.

Stationary Wavelet Transform and AdaBoost with SVM Based Pathological Brain Detection in MRI Scanning

This paper presents an automatic classification system for segregating pathological brain from normal brains in magnetic resonance imaging scanning. The proposed system employs contrast limited adaptive histogram equalization scheme to enhance the diseased region in brain MR images. Two-dimensional stationary wavelet transform is harnessed to extract features from the preprocessed images. The feature vector is constructed using the energy and entropy values, computed from the level- 2 SWT coefficients. Then, the relevant and uncorrelated features are selected using symmetric uncertainty ranking filter. Subsequently, the selected features are given input to the proposed AdaBoost with support vector machine classifier, where SVM is used as the base classifier of AdaBoost algorithm. To validate the proposed system, three standard MR image datasets, Dataset-66, Dataset-160, and Dataset- 255 have been utilized. The 5 runs of k-fold stratified cross validation results indicate the suggested scheme offers better performance than other existing schemes in terms of accuracy and number of features. The proposed system earns ideal classification over Dataset-66 and Dataset-160; whereas, for Dataset- 255, an accuracy of 99.45% is achieved.

Detection of unilateral hearing loss by Stationary Wavelet Entropy.

AIM Sensorineural hearing loss is correlated to massive neurological or psychiatric disease. MATERIALS T1-weighted volumetric images were acquired from fourteen subjects with right-sided hearing loss (RHL), fifteen subjects with left-sided hearing loss (LHL), and twenty healthy controls (HC). METHOD We treated a three-class classification problem: HC, LHL, and RHL. Stationary wavelet entropy was employed to extract global features from magnetic resonance images of each subject. Those stationary wavelet entropy features were used as input to a single-hidden layer feedforward neuralnetwork classifier. RESULTS The 10 repetition results of 10-fold cross validation show that the accuracies of HC, LHL, and RHL are 96.94%, 97.14%, and 97.35%, respectively. CONCLUSION Our developed system is promising and effective in detecting hearing loss.

Pathological brain detection using curvelet features and least squares SVM

This paper aims at developing an automatic pathological brain detection system (PBDS) to assist radiologists in identifying brain diseases correctly in less time. Magnetic resonance imaging (MRI) has the potential to provide better information about the brain soft tissues and hence MR images have been incorporated in the proposed system. Fifty largest coefficients are selected from each sub-band of a level-5 fast discrete curvelet transform (FDCT) to serve as a feature set for each image. To reduce the size of the feature set, principal component analysis (PCA) has been harnessed. Subsequently, least squares SVM (LS-SVM) with three different kernels are utilized to segregate the images as healthy or pathological. The proposed system has been validated on three benchmark datasets and a 10 ×k-fold stratified cross validation (SCV) test has been performed. It indicates that the proposed system “FDCT + PCA + LS-SVM + RBF” achieves better performance than not only two other systems having linear and polynomial kernel but also 22 existing methods. In addition, the suggested system requires only six features which are computationally economical for a practical use.

Automated pathological brain detection system: A fast discrete curvelet transform and probabilistic neural network based approach

Abstract Computer-aided diagnosis (CAD) systems have drawn attention of researchers for arriving at qualitative and faster clinical decisions, and hence has become one of the most important directions of research. In this paper, we propose an efficient CAD system to classify pathological and healthy brains using brain MR images. The suggested pathological brain detection system (PBDS) has the ability to help radiologists to initiate the corrective measures for treating the ailing patients at an early stage. The proposed scheme uses a simplified pulse-coupled neural network (SPCNN) for the region of interest (ROI) segmentation and fast discrete curvelet transform (FDCT) for feature extraction. Subsequently, PCA+LDA approach is harnessed for feature dimensionality reduction and finally probabilistic neural network (PNN) is applied for classification. The scheme is validated on various standard datasets and compared with existing competent schemes with respect to classification accuracy and number of features. The statistical set up is kept similar as reported in the recent literature to derive an unbiased analysis. Experimental results demonstrate that the suggested scheme yields higher accuracy as compared to others with considerably less number of features. The number of parameters need to be tuned at different stages are significantly less in contrast to existing schemes. Further, PNN used has a simple network structure and offers faster learning speed. Therefore, the proposed scheme can effectively detect pathological brain in real-time and hence has a potential to be installed on medical robots.

Classification of brain MR images using discrete wavelet transform and random forests

Development of computer-aided diagnosis (CAD) systems for early detection of the pathological brain is essential to save medical resources. In recent years, a variety of techniques have been proposed to upgrade the systems performance. In this paper, a new automatic CAD system for brain magnetic resonance (MR) image classification is proposed. The method utilizes two-dimensional discrete wavelet transform to extract features from the MR images. The dimension of the features have been reduced using principal component analysis (PCA) and linear discriminant analysis (LDA), to obtain the more significant features. Finally, the reduced set of features are applied to the random forests classifier to determine the normal or pathological brain. A standard dataset, Dataset-255 of 255 images (35 normal and 220 pathological) is used for the validation of the proposed scheme. To improve the generalization capability of the scheme, 5-fold stratified cross-validation procedure is utilized. The results of the experiments reveal that the proposed scheme is superior to other state-of-the-art techniques in terms of classification accuracy with substantially reduced number of features.

An efficient edge detection technique by two dimensional rectangular cellular automata

This paper proposes a new pattern of two dimensional cellular automata linear rules that are used for efficient edge detection of an image. Since cellular automata is inherently parallel in nature, it has produced desired output within a unit time interval. We have observed four linear rules among 29 total linear rules of a rectangular cellular automata in adiabatic or reflexive boundary condition that produces an optimal result. These four rules are directly applied once to the images and produced edge detected output. We compare our results with the existing edge detection algorithms and found that our results shows better edge detection with an enhancement of edges.

Least squares SVM approach for abnormal brain detection in MRI using multiresolution analysis

Developing automatic and accurate computer-aided diagnosis (CAD) systems for detecting brain disease in magnetic resonance imaging (MRI) are of great importance in recent years. These systems help the radiologists in accurate interpretation of brain MR images and also substantially reduce the time needed for it. In this paper, a new system for abnormal brain detection is presented. The proposed method employs a multiresolution approach (discrete wavelet transform) to extract features from the MR images. Kernel principal component analysis (KPCA) is harnessed to reduce the dimension of the features, with the goal of obtaining the discriminant features. Subsequently, a new version of support vector machine (SVM) with low computational cost, called least squares SVM (LS-SVM) is utilized to classify brain MR images as normal or abnormal. The proposed scheme is validated on a dataset of 90 images (18 normal and 72 abnormal). A 6-fold stratified cross-validation procedure is implemented and the results of the experiments indicate that the proposed scheme outperforms other competent schemes in terms of classification accuracy with relatively small number of features.

Salt and Pepper Noise Reduction Schemes Using Cellular Automata

Two filters in the light of two-dimensional Cellular Automata (CA) are presented in this paper for salt and pepper noise reduction of an image. The design of a parallel algorithm to remove noise from corrupted images is a demanded approach now, so we utilize the idea of cellular automata to cater this need. The filters are mainly designed according to the neighborhood structure of a cell with different boundary conditions. The performances of the proposed filters with that of existing filters are evaluated in terms of peak signal-to-noise ratio (PSNR) values and it has been observed that the proposed filters are extremely promising for noise reduction of an image contaminated by salt and pepper noise. The primary point of interest in utilizing these proposed filters is; it preserves more image details in expense of noise suppression.

Development of pathological brain detection system using Jaya optimized improved extreme learning machine and orthogonal ripplet-II transform

Pathological brain detection systems (PBDSs) have drawn much attention from researchers over the past two decades because of their significance in taking correct clinical decisions. In this paper, an efficient PBDS based on MR images is introduced that markedly improves the recent results. The proposed system makes use of contrast limited adaptive histogram equalization (CLAHE) and orthogonal discrete ripplet-II transform (O-DR2T) with degree 2 to enhance the quality of the input MR images and extract the features respectively. Subsequently, relevant features are obtained using PCA+LDA approach. Finally, a novel learning algorithm called IJaya-ELM is proposed that combines improved Jaya algorithm (IJaya) and extreme learning machine (ELM) for segregation of MR images as pathological or healthy. The improved Jaya algorithm is utilized to optimize the input weights and hidden biases of single-hidden-layer feedforward neural networks (SLFN), whereas one analytical method is used for determining the output weights. The proposed algorithm performs optimization according to both the root mean squared error (RMSE) and the norm of the output weights of SLFNs. Extensive experiments are carried out using three benchmark datasets and the results are compared against other competent schemes. The experimental results demonstrate that the proposed scheme brings potential improvements in terms of classification accuracy and number of features. Moreover, the proposed IJaya-ELM classifier achieves higher accuracy and obtains compact network architecture compared to conventional ELM and BPNN classifier.