W. Mansor

Classification of Infant Cries with Asphyxia Using Multilayer Perceptron Neural Network

Asphyxia occurs in infants with neurological level disturbance, which is found to affect sound of cry produced by infants. The infant cry signals with asphyxia have distinct patterns which can be recognized with pattern classifiers such as Artificial Neural Network (ANN). This study investigates the performance of the Multilayer Perceptron (MLP) classifier in discriminating between healthy and infants with asphyxia from their cries, of ages from zero to seven months old, with an input feature reduction algorithm, Orthogonal Lest Square (OLS) analysis, in contrast to direct selection. The infant cry waveform served as input to Mel Frequency Cepstrum (MFC) analysis for feature extraction. The MLP classifier performance was examined with different combination in number of coefficients, filter bank and hidden nodes. It is found that the OLS algorithm is effective in enhancing the accuracy of MLP classifier while reducing the computation load. Both the average and highest MLP classification accuracies with coefficients being ranked by OLS algorithm have consistently displayed better score than those by direct selection. The highest MLP classification accuracy of 94% is obtained with 40 filter banks, 12 highly ranked MFC coefficients and 15 hidden nodes.

EOG signal detection for home appliances activation

Eye movement is the most common way for paralysis patients to communicate. Eye movement can be used by the paralysis patients and armless persons to perform simple tasks, for examples to activate a system in order to get attention or to switch home appliances on and off when they are alone. This study investigates the reliability of EOG signals for activating home appliances. A replica of a living room with a fixed location of the television and a switch was set up and the EOG signals were recorded when the subjects were sitting and standing at various positions. Various distances between the subject and the television and gaze angles were considered to obtain optimum EOG signals. An algorithm to detect the occurrence of the EOG signals was developed and its performance was examined. It was found that the proposed eye movement technique could produce clear EOG signals which are suitable for activating home appliances and the success percentage of the algorithm is 92%.

An analysis of EEG signal generated from grasping and writing

Electroencephalogram consists of hand movement information that can be extracted using suitable digital signal processing techniques. In this study, the EEG signals generated from hand grasping and writing were recorded from 4 channels; C3, C4, P3 and P4 and filtered using band pass filter with frequency range of 8 Hz to 30 Hz. The signal was then analysed using Fast Fourier Transform. Analysis of EEG signals showed that both hand grasping and writing produced signals with beta frequency.

Performance of Combined Support Vector Machine and Principal Component Analysis in recognizing infant cry with asphyxia

Combined Support Vector Machine (SVM) and Principal Component Analysis (PCA) was used to recognize the infant cries with asphyxia. SVM classifier based on features selected by the PCA was trained to differentiate between pathological and healthy cries. The PCA was applied to reduce dimensionality of the vectors that serve as inputs to the SVM. The performance of the SVM utilizing linear and RBF kernel was examined. Experimental results showed that SVM with RBF kernel yields good performance. The classification accuracy in classifying infant cry with asphyxia using the SVM-PCA is 95.86%.

Binary Particle Swarm Optimization for selection of features in the recognition of infants cries with asphyxia

The infant cry signals with asphyxia have distinct patterns which can be recognized using pattern classifiers such as Artificial Neural Network (ANN). This study investigates the effect of selecting infant cry features using the Binary Particle Swarm Optimization on the performance of Multilayer Perceptron (MLP) classifier in discriminating between healthy and infants with asphyxia from cry signals. The feature extraction process was performed by MFCC analysis. The MLP classifier performance was examined using various combination of number of coefficients. It was found that the BPSO helps to enhance the classification accuracy of MLP classifier while reducing the computational load. The highest MLP classification accuracy achieved was 95.07%, which was obtained when 26 MFCC filter banks, 14 selected MFC coefficients and 5 hidden nodes were used.

Review of brain computer interface application in diagnosing dyslexia

Dyslexia is one of brain disorders which needs to be detected at an early stage to allow the children to master the basic and to avoid damage to self esteem and self-confidence. In this paper, treatment of dyslexia and researches on dyslexia diagnosis are discussed. Neuro-feedback has high potential to diagnose dyslexia. It has been proven that neuro-feedback is able to improve spelling disorder. Therefore, investigation on the performance of neuro-feedback in diagnosing and treating reading disorder should be carried out.

Classification of infant cries with hypothyroidism using Multilayer Perceptron neural network

Hypothyroidism occurs in infants with insufficient production of hormones by the thyroid gland. The cry signals of babies with hypothyroidism have distinct patterns which can be recognized with pattern classifiers such as Multilayer Perceptron (MLP) artificial neural network. This study investigates the performance of the MLP in discriminating between healthy infants and infants suffering from hypothyroidism based on their cries. The infant cries were first divided into one second segments, and important features were extracted using Mel Frequency Cepstrum Coefficient (MFCC) analysis. Two methods were then used to select which MFCC coefficients to be used as features for the MLP: direct selection or Fishers Ratio analysis (F-ratio analysis). Their performances were compared with experimental results showing that MLP was able to accurately distinguish between the two cases. The classification performance of MLP trained with F-Ratio analysis is found to be better compared to direct selection method.

Classification of salivary based NS1 from Raman Spectroscopy with support vector machine

Non-Structural Protein 1 (NS1) antigen has been recognized as a biomarker for diagnosis of flavivirus viral infections at early stage. Surface Enhanced Raman Spectroscopy (SERS) is an optical technique capable of detecting up to a single molecule. Our previous work has established the Raman fingerprint of NS1 with gold as substrate. Our current study aims to classify NS1 infected saliva samples from healthy samples, a first ever attempt. Saliva samples from healthy subjects, NS1 protein and NS1-saliva mixture samples were analyzed using SERS. The SERS spectra were then pre-processed prior to classification with support vector machine (SVM). NS1-saliva mixture at concentration of 10ppm, 50ppm and 100ppm were examined. Performance of SVM classifier with linear, polynomial and radial basis function (RBF) kernels were compared, in term of accuracy, sensitivity, and specificity. From the results, it can be concluded that SVM classifier is able to classify the samples into NS1 infected samples and normal saliva samples. Of the three kernels, performance in using polynomial and RBF kernel is found surpassing the linear kernel. The best performance is attained with RBF kernel with accuracy of [97.1% 93.4% 81.5%] for 100ppm, 50ppm and 10ppm respectively.

Raman molecular fingerprint of non-structural protein 1 in phosphate buffer saline with gold substrate

SERS is a form of Raman spectroscopy that is enhanced with nano-sensing chip as substrate. It can yield distinct biochemical fingerprint for molecule of solids, liquids and gases. Vice versa, it can be used to identify unknown molecule. It has further advantage of being non-invasive, non-contact and cheap, as compared to other existing laboratory based techniques. NS1 has been clinically accepted as an alternative biomarker to IgM in diagnosing viral diseases carried by virus of flaviviridae. Its presence in the blood serum at febrile stage of the flavivirus infection has been proven. Being an antigen, it allows early detection that can help to reduce the mortality rate. This paper proposes SERS as a technique for detection of NS1 from its scattering spectrum. Contribution from our work so far has never been reported. From our experiments, it is found that NS1 protein is Raman active. Its spectrum exhibits five prominent peaks at Raman shift of 548, 1012, 1180, 1540 and 1650cm-1. Of these, peak at 1012cm-1 scales the highest intensity. It is singled out as the peak to fingerprint the NS1 protein. This is because its presence is verified by the ring breathing vibration of the benzene ring structure side chain molecule. The characteristic peak is found to vary in proportion to concentration. It is found that for a 99% change in concentration, a 96.7% change in intensity is incurred. This yields a high sensitivity of about one a.u. per ppm. Further investigation from the characterization graph shows a correlation coefficient of 0.9978 and a standard error estimation of 0.02782, which strongly suggests a linear relationship between the concentration and characteristic peak intensity of NS1. Our finding produces favorable evidence to the use of SERS technique for detection of NS1 protein for early detection of flavivirus infected diseases with gold substrate.

Optimized Support Vector Machine for classifying infant cries with asphyxia using Orthogonal Least Square

This paper investigates the effect of optimizing Support Vector Machine, with linear and RBF kernels, on its performance in classifying asphyxiated infant cries, with Orthogonal Least Square. Mel Frequency Cepstrum analysis first extracts feature from the infant cry signals. The extracted features are then ranked in accordance to its error reduction ratio with OLS. SVM with linear and RBF kernel then classify the asphyxiated infant cry from the optimized and non-optimized input feature vector. The classification accuracy and support vector number are used to gauge the performance. Experimental result shows that for both kernels, the OLS-optimized SVM achieve equally high classification accuracy with lower support vector number than the non-optimized one. It is also found that the OLS-SVM with RBF kernel outperformed all other methods with classification accuracy of 93.16% and support vector number of 266.2.