M.U.S. Perera

Bio: M.U.S. Perera is an academic researcher from Informatics Institute of Technology. The author has contributed to research in topics: Wavelet transform & Electroencephalography. The author has an hindex of 6, co-authored 10 publications receiving 82 citations.

Optical flow based obstacle avoidance for the visually impaired

Abstract: Vision is a vital cue for human navigation. Thus, visually impaired people encounter many challenges in day-today travelling. Identifying and avoiding obstacles in the environment is the most crucial among them. To empower blind navigation, numerous electronic travel aids were created in the past few decades, by using various obstacle sensing technologies such as sonar, infrared, and stereo vision. However, optical flow estimations based navigation, which is heavily used by insects and experimented in the field of robotics, has not been used in them. This project aimed to evaluate the potential optical flow estimation based techniques has in guiding a visually impaired person to avoid obstacles using auditory and tactile feedback. To demonstrate the researched core concepts, a prototype consisting of a virtual reality world was designed and developed. It also employs an existing optical flow algorithm for motion estimation, other image processing techniques, speech synthesis for auditory feedback and embedded programming for tactile feedback. The modular design of the prototype enables it to be used either in simulation mode, or as a standalone application in a real world environment. This work demonstrates the attractive possibilities of using optical flow estimations for visually impaired navigation.

Intelligent brain hemorrhage diagnosis using artificial neural networks

Abstract: Brain hemorrhage is a type of stroke which is caused by an artery in the brain bursting and causing bleeding in the surrounded tissues. Diagnosing brain hemorrhage, which is mainly through the examination of a CT scan enables the accurate prediction of disease and the extraction of reliable and robust measurement for patients in order to describe the morphological changes in the brain as the recovery progresses. Though a lot of research on medical image processing has been done, still there is opportunity for further research in the area of brain hemorrhage diagnosis due to the low accuracy level in the current methods and algorithms, coding complexity of the developed approaches, impracticability in the real environment, and lack of other enhancements which may make the system more interactive and useful. Additionally many of the existing approaches address the diagnosis of a limited no of brain hemorrhage types. This project investigates the possibility of diagnosing brain hemorrhage using an image segmentation of CT scan images using watershed method and feeding of the appropriate inputs extracted from the brain CT image to an artificial neural network for classification. The output generated as the type of brain hemorrhages, can be used to verify expert diagnosis and also as a learning tool for trainee radiologists to minimize errors in current methods. The prototype developed using Matlab can help medical students to practice the related concepts they learn using an image guide with examples for surgeries and surgical simulation. System was evaluated by the domain experts, like radiologists, intended users such as medical students as well as by technical experts. The prototype developed was successful since it was being evaluated as credible, innovative and useful software for the students in the field of radiology while 100% of the evaluators mentioned the diagnosis accuracy is acceptable.

Forecasting epileptic seizures using EEG signals, wavelet transform and artificial neural networks

Abstract: Electroencephalograms (EEG) are signal records of electrical activity of brain neurons EEG, which is a compulsive tool/used for diagnosing neurological diseases such as epilepsy, besides of techniques such as magnetic resonance and brain tomography (BT) that are used for diagnosing structural brain disorders This paper describes a novel approach for forecasting epileptic seizure activity, by classifying these EEG signals The decision making consists of two stages; initially the signal features are extracted by applying wavelet transform (WT) and then an artificial neural network (ANN) model, which is a supervised learning-based algorithm classifier, used for signal classification Wavelet transform is an effective tool for analysis of transient events in non-stationary signals, such as EEGs The performance of the ANN classifier is evaluated in terms of sensitivity, specificity and classification accuracy The obtained classification

Intelligent emotion recognition system using brain signals (EEG)

Abstract: Human emotion is a complex and psycho physiological state of mind which can be expressed as positive or negative reactions to external and internal stimuli. Typical communication channels that indicate emotions are voice and facial expressions. People who are paralyzed or have other severe movement disorders have no way of expressing their emotions thereby forming a wide communication rift between them and the outside world. Communication through eye tracking is one of the alternative ways of giving such disabled patients to interact with the outside world. This paper investigates the possibility of recognizing emotions using signal processing of Electroencephalography using discrete wavelet transform and feeding appropriate values to an adaptive neuro fuzzy inference system for classification. The system enables severely disabled as well as able users to interact with the system using eye movement in order to respond to detected emotion. The solution can be used to detect emotions of motor disabled people and provision a means of communication; also it is a learning tool for trainee neurologists. The prototype was built using Matlab successfully and it was evaluated by experts and the intended users very creditably stating it as a useful software for disabled people.

Focal epileptic seizure forecasting in EEG signals using wavelet transform and artificial neural networks

Abstract: Seizure prediction has become a major field of neurological research, because of the suggestion of recent research that electrophysiological changes develop minutes to hours, before the actual clinical onset in focal epileptic seizures. This paper describes a novel approach for forecasting focal epileptic seizures, by applying statistical analysis methods and classifying, the features derived from intracranial Electroencephalographic (EEG) recordings, of brain activity. The decision making consists of two stages; initially the signal features are extracted by applying wavelet transform (WT) and then an artificial neural network (ANN) model, which is a supervised learning-based algorithm classifier, used for signal classification. Wavelet transform is an effective tool for analysis of transient events in nonstationary signals, such as EEGs. The performance of the ANN classifier is evaluated in terms of sensitivity, specificity and classification accuracy. The obtained classification accuracy confirms that the proposed scheme has potential in classifying EEG signals.

Statistical Pattern Recognition

Abstract: This chapter introduces the subject of statistical pattern recognition (SPR). It starts by considering how features are defined and emphasizes that the nearest neighbor algorithm achieves error rates comparable with those of an ideal Bayes’ classifier. The concepts of an optimal number of features, representativeness of the training data, and the need to avoid overfitting to the training data are stressed. The chapter shows that methods such as the support vector machine and artificial neural networks are subject to these same training limitations, although each has its advantages. For neural networks, the multilayer perceptron architecture and back-propagation algorithm are described. The chapter distinguishes between supervised and unsupervised learning, demonstrating the advantages of the latter and showing how methods such as clustering and principal components analysis fit into the SPR framework. The chapter also defines the receiver operating characteristic, which allows an optimum balance between false positives and false negatives to be achieved.

Epileptic Seizure Classification of EEG Time-Series Using Rational Discrete Short-Time Fourier Transform

Abstract: A system for epileptic seizure detection in electroencephalography (EEG) is described in this paper. One of the challenges is to distinguish rhythmic discharges from nonstationary patterns occurring during seizures. The proposed approach is based on an adaptive and localized time–frequency representation of EEG signals by means of rational functions. The corresponding rational discrete short-time Fourier transform (DSTFT) is a novel feature extraction technique for epileptic EEG data. A multilayer perceptron classifier is fed by the coefficients of the rational DSTFT in order to separate seizure epochs from seizure-free epochs. The effectiveness of the proposed method is compared with several state-of-art feature extraction algorithms used in offline epileptic seizure detection. The results of the comparative evaluations show that the proposed method outperforms competing techniques in terms of classification accuracy. In addition, it provides a compact representation of EEG time-series.

Face recognition using Principal Component Analysis

Abstract: The strategy of face recognition involves the examination of facial features in a picture, recognizing those features and matching them to 1 of the many faces in the database. There are lots of algorithms effective at performing face recognition, such as for instance: Principal Component Analysis, Discrete Cosine Transform, 3D acceptance methods, Gabor Wavelets method etc. This work has centered on Principal Component Analysis (PCA) method for face recognition in an efficient manner. There are numerous issues to take into account whenever choosing a face recognition method. The main element is: Accuracy, Time limitations, Process speed and Availiability. With one of these in minds PCA way of face recognition is selected because it is really a simplest and easiest approach to implement, extremely fast computation time. PCA (Principal Component Analysis) is an activity that extracts the absolute most relevant information within a face and then tries to construct a computational model that best describes it.

Epileptic seizure detection in long-term EEG records using sparse rational decomposition and local Gabor binary patterns feature extraction

Abstract: In this paper, we address the problem of off-line supervised detection of epileptic seizures in long-term Electroencephalography (EEG) records. A novel feature extraction method is proposed based on the sparse rational decomposition and the Local Gabor Binary Patterns (LGBP). Namely, we decompose the channels of the EEG record into 8 sparse rational components using a group of optimal coefficients. Then, a modified 1D LGBP operator is applied, which is followed by downsampling of the data. The width of the largest LGBPs is finally computed for all the 8 rational components and the 23 channels of the EEG record. Hence, we characterize seizure patterns of one-second-long EEG epochs by 23 8 features. The effectiveness of the proposed feature extraction method is assessed using different classifiers which are trained with 25% of early EEG records of each patient. We performed an extensive comparative study over 163h of EEG recordings from the CHB-MIT Scalp EEG database. The experiments show that the proposed technique outperforms other dedicated techniques by achieving the overall sensitivity of 70.4% and the overall specificity of 99.1%, in the patient-specific detection of epileptic EEG epochs. Moreover, it detects onset of seizures with the overall sensitivity of 91.13% and false alarms per hour rate of 0.35, on average.