Detection of K-complexes in EEG waveform images using faster R-CNN and deep transfer learning.

BACKGROUND: The electroencephalography (EEG) signal carries important information about the electrical activity of the brain, which may reveal many pathologies. This information is carried in certain waveforms and events, one of which is the K-complex. It is used by neurologists to diagnose neurophysiologic and cognitive disorders as well as sleep studies. Existing detection methods largely depend on tedious, time-consuming, and error-prone manual inspection of the EEG waveform. METHODS: In this paper, a highly accurate K-complex detection system is developed. Based on multiple convolutional neural network (CNN) feature extraction backbones and EEG waveform images, a regions with faster regions with convolutional neural networks (Faster R-CNN) detector was designed, trained, and tested. Extensive performance evaluation was performed using four deep transfer learning feature extraction models (AlexNet, ResNet-101, VGG19 and Inceptionv3). The dataset was comprised of 10948 images of EEG waveforms, with the location of the K-complexes included as separate text files containing the bounding boxes information. RESULTS: The Inceptionv3 and VGG19-based detectors performed consistently high (i.e., up to 99.8% precision and 0.2% miss rate) over different testing scenarios, in which the number of training images was varied from 60% to 80% and the positive overlap threshold was increased from 60% to 90%. CONCLUSIONS: Our automated method appears to be a highly accurate automatic K-complex detection in real-time that can aid practitioners in speedy EEG inspection.

Using deep transfer learning to detect scoliosis and spondylolisthesis from x-ray images.

Recent years have witnessed wider prevalence of vertebral column pathologies due to lifestyle changes, sedentary behaviors, or injuries. Spondylolisthesis and scoliosis are two of the most common ailments with an incidence of 5% and 3% in the United States population, respectively. Both of these abnormalities can affect children at a young age and, if left untreated, can progress into severe pain. Moreover, severe scoliosis can even lead to lung and heart problems. Thus, early diagnosis can make it easier to apply remedies/interventions and prevent further disease progression. Current diagnosis methods are based on visual inspection by physicians of radiographs and/or calculation of certain angles (e.g., Cobb angle). Traditional artificial intelligence-based diagnosis systems utilized these parameters to perform automated classification, which enabled fast and easy diagnosis supporting tools. However, they still require the specialists to perform error-prone tedious measurements. To this end, automated measurement tools were proposed based on processing techniques of X-ray images. In this paper, we utilize advances in deep transfer learning to diagnose spondylolisthesis and scoliosis from X-ray images without the need for any measurements. We collected raw data from real X-ray images of 338 subjects (i.e., 188 scoliosis, 79 spondylolisthesis, and 71 healthy). Deep transfer learning models were developed to perform three-class classification as well as pair-wise binary classifications among the three classes. The highest mean accuracy and maximum accuracy for three-class classification was 96.73% and 98.02%, respectively. Regarding pair-wise binary classification, high accuracy values were achieved for most of the models (i.e., > 98%). These results and other performance metrics reflect a robust ability to diagnose the subjects' vertebral column disorders from standard X-ray images. The current study provides a supporting tool that can reasonably help the physicians make the correct early diagnosis with less effort and errors, and reduce the need for surgical interventions.

Non-Contact Spirometry Using a Mobile Thermal Camera and AI Regression.

Non-contact physiological measurements have been under investigation for many years, and among these measurements is non-contact spirometry, which could provide acute and chronic pulmonary disease monitoring and diagnosis. This work presents a feasibility study for non-contact spirometry measurements using a mobile thermal imaging system. Thermal images were acquired from 19 subjects for measuring the respiration rate and the volume of inhaled and exhaled air. A mobile application was built to measure the respiration rate and export the respiration signal to a personal computer. The mobile application acquired thermal video images at a rate of nine frames/second and the OpenCV library was used for localization of the area of interest (nose and mouth). Artificial intelligence regressors were used to predict the inhalation and exhalation air volume. Several regressors were tested and four of them showed excellent performance: random forest, adaptive boosting, gradient boosting, and decision trees. The latter showed the best regression results, with an R-square value of 0.9998 and a mean square error of 0.0023. The results of this study showed that non-contact spirometry based on a thermal imaging system is feasible and provides all the basic measurements that the conventional spirometers support.

Detection of COVID-19 from Chest X-ray Images Using Deep Convolutional Neural Networks.

The COVID-19 global pandemic has wreaked havoc on every aspect of our lives. More specifically, healthcare systems were greatly stretched to their limits and beyond. Advances in artificial intelligence have enabled the implementation of sophisticated applications that can meet clinical accuracy requirements. In this study, customized and pre-trained deep learning models based on convolutional neural networks were used to detect pneumonia caused by COVID-19 respiratory complications. Chest X-ray images from 368 confirmed COVID-19 patients were collected locally. In addition, data from three publicly available datasets were used. The performance was evaluated in four ways. First, the public dataset was used for training and testing. Second, data from the local and public sources were combined and used to train and test the models. Third, the public dataset was used to train the model and the local data were used for testing only. This approach adds greater credibility to the detection models and tests their ability to generalize to new data without overfitting the model to specific samples. Fourth, the combined data were used for training and the local dataset was used for testing. The results show a high detection accuracy of 98.7% with the combined dataset, and most models handled new data with an insignificant drop in accuracy.

Computer-aided identification of degenerative neuromuscular diseases based on gait dynamics and ensemble decision tree classifiers.

This study proposes a reliable computer-aided framework to identify gait fluctuations associated with a wide range of degenerative neuromuscular disease (DNDs) and health conditions. Investigated DNDs included amyotrophic lateral sclerosis (ALS), Parkinson's disease (PD), and Huntington's disease (HD). We further performed a statistical and classification comparison elucidating the discriminative capability of different gait signals, including vertical ground reaction force (VGRF), stride duration, stance duration, and swing duration. Feature representation of these gait signals was based on statistical amplitude quantification using the root mean square (RMS), variance, kurtosis, and skewness metrics. We investigated various decision tree (DT) based ensemble methods such as bagging, adaptive boosting (AdaBoost), random under-sampling boosting (RUSBoost), and random subspace to tackle the challenge of multi-class classification. Experimental results showed that AdaBoost ensembling provided a 6.49%, 0.78%, 2.31%, and 2.72% prediction rate improvement for the VGRF, stride, stance, and swing signals, respectively. The proposed approach achieved the highest classification accuracy of 99.17%, sensitivity of 98.23%, and specificity of 99.43%, using the VGRF-based features and the adaptive boosting classification model. This work demonstrates the effective capability of using simple gait fluctuation analysis and machine learning approaches to detect DNDs. Computer-aided analysis of gait fluctuations provides a promising advent to enhance clinical diagnosis of DNDs.

A dataset of lung sounds recorded from the chest wall using an electronic stethoscope.

The advancement of stethoscope technology has enabled high quality recording of patient sounds. We used an electronic stethoscope to record lung sounds from healthy and unhealthy subjects. The dataset includes sounds from seven ailments (i.e., asthma, heart failure, pneumonia, bronchitis, pleural effusion, lung fibrosis, and chronic obstructive pulmonary disease (COPD)) as well as normal breathing sounds. The dataset presented in this article contains the audio recordings from the examination of the chest wall at various vantage points. The stethoscope placement on the subject was determined by the specialist physician performing the diagnosis. Each recording was replicated three times corresponding to various frequency filters that emphasize certain bodily sounds. The dataset can be used for the development of automated methods that detect pulmonary diseases from lung sounds or identify the correct type of lung sound. The same methods can also be applied to the study of heart sounds.

Convolutional and recurrent neural networks for the detection of valvular heart diseases in phonocardiogram recordings.

Valvular heart diseases (VHD) are one of the major causes of cardiovascular diseases that are having high mortality rates worldwide. The early diagnosis of VHD prevents the development of cardiac diseases and allows for optimum medication. Despite of the ability of current gold standards in identifying VHD, they still lack the required accuracy and thus, several cases go misdiagnosed. In this vein, a study is conducted herein to investigate the efficiency of deep learning models in identifying VHD through phonocardiography (PCG) recordings. PCG heart sounds were obtained from an open-access data-set representing normal heart sounds along with four major VHD; namely aortic stenosis (AS), mitral stenosis (MS), mitral regurgitation (MR), and mitral valve prolapse (MVP). A total of 1,000 patients were involved in the study with 200 recordings for each class. All recordings were initially trimmed to have 9,600 samples ensuring their coverage of at least 1 cardiac cycle. In addition, they were pre-processed by applying maximal overlap discrete wavelet transform (MODWT) smoothing algorithm and z-score normalization. The neural network architecture was designed to reduce the complexity often found in literature and consisted of a combination of convolutional neural networks (CNN) and recurrent neural networks (RNN) based on Bi-directional long short-term memory (BiLSTM). The model was trained and tested following a k-fold cross-validation scheme of 10-folds utilizing the CNN-BiLSTM network as well as the CNN and BiLSTM, individually. The highest performance was achieved using the CNN-BiLSTM network with an overall Cohen's kappa, accuracy, sensitivity, and specificity of 97.87%, 99.32%, 98.30%, and 99.58%, respectively. In addition, the model had an average area under the curve (AUC) of 0.998. Furthermore, the performance of the model was assessed on the PhysioNet/Computing in Cardiology 2016 challenge data-set and reached an overall accuracy of 87.31% with an AUC of 0.900. This study paves the way towards implementing deep learning models in VHD identification under clinical settings to assist clinicians in decision making and prevent many cases from cardiac abnormalities development.

A convolutional neural network for the screening and staging of diabetic retinopathy.

Diabetic retinopathy (DR) is a serious retinal disease and is considered as a leading cause of blindness in the world. Ophthalmologists use optical coherence tomography (OCT) and fundus photography for the purpose of assessing the retinal thickness, and structure, in addition to detecting edema, hemorrhage, and scars. Deep learning models are mainly used to analyze OCT or fundus images, extract unique features for each stage of DR and therefore classify images and stage the disease. Throughout this paper, a deep Convolutional Neural Network (CNN) with 18 convolutional layers and 3 fully connected layers is proposed to analyze fundus images and automatically distinguish between controls (i.e. no DR), moderate DR (i.e. a combination of mild and moderate Non Proliferative DR (NPDR)) and severe DR (i.e. a group of severe NPDR, and Proliferative DR (PDR)) with a validation accuracy of 88%-89%, a sensitivity of 87%-89%, a specificity of 94%-95%, and a Quadratic Weighted Kappa Score of 0.91-0.92 when both 5-fold, and 10-fold cross validation methods were used respectively. A prior pre-processing stage was deployed where image resizing and a class-specific data augmentation were used. The proposed approach is considerably accurate in objectively diagnosing and grading diabetic retinopathy, which obviates the need for a retina specialist and expands access to retinal care. This technology enables both early diagnosis and objective tracking of disease progression which may help optimize medical therapy to minimize vision loss.

A Comprehensive Framework for Differentiating Autism Spectrum Disorder From Neurotypicals by Fusing Structural MRI and Resting State Functional MRI.

Autism spectrum disorder is a neurodevelopmental disorder characterized by impaired social abilities and communication difficulties. The golden standard for autism diagnosis in research rely on behavioral features, for example, the autism diagnosis observation schedule, the Autism Diagnostic Interview-Revised. In this study we introduce a computer-aided diagnosis system that uses features from structural MRI (sMRI) and resting state functional MRI (fMRI) to help predict an autism diagnosis by clinicians. The proposed system is capable of parcellating brain regions to show which areas are most likely affected by autism related abnormalities and thus help in targeting potential therapeutic interventions. When tested on 18 data sets (n = 1060) from the ABIDE consortium, our system was able to achieve high accuracy (sMRI 0.75-1.00; fMRI 0.79-1.00), sensitivity (sMRI 0.73-1.00; fMRI 0.78-1.00), and specificity (sMRI 0.78-1.00; fMRI 0.79-1.00). The proposed system could be considered an important step toward helping physicians interpret results of neuroimaging studies and personalize treatment options. To the best of our knowledge, this work is the first to combine features from structural and functional MRI, use them for personalized diagnosis and achieve high accuracies on a relatively large population.

Neonatal sleep stage identification using long short-term memory learning system.

Neonatal sleep analysis at the neonatal intensive care units (NICU) is critical for the diagnosis of any brain growth risks during the early stages of life. In this paper, an investigation is carried out on the use of a long short-term memory (LSTM) learning system in automatic sleep stage scoring in neonates. The developed algorithm automatically classifies sleep stages based on inputs from a single channel EEG recording. Up to this date, only a single study have developed an approach for automatic sleep stage scoring in neonatal sleep signals using deep neural network (DNN). A total of 5095 sleep stages signals acquired from EEG recordings of the University of Pittsburgh are used in this study. The sleep stages are annotated by a medical doctor from the Pediatric Neurology Department of Case Western Reserve University for three neonatal sleep stages including the awake (W), active sleep (AS), and quiet sleep (QS) stages on every 60-s epoch. The signals are pre-processed through normalization and filtering. The resulted signals are divided following 4-, 6-, and 10-fold cross-validation schemes. The training and classification process is done using a bi-directional LSTM network classifier built with pre-defined training parameters. At the end, the developed algorithm is evaluated along with a complete summary table that reports the results of this study and other state-of-the-art studies. The current study achieved high levels of Cohen's kappa (κ), accuracy, and F1 score with 91.37%, 96.81%, and 94.43%, respectively. Based on the confusion matrix, the overall true positives percentage reached 95.21%. The developed algorithm gave promising results in automatic sleep stage scoring in neonatal sleep signals. Future work include LSTM architecture and training parameters improvements to enhance the overall accuracy of the classifier.

A multimodal computer-aided diagnostic system for precise identification of renal allograft rejection: Preliminary results.

PURPOSE: Early assessment of renal allograft function post-transplantation is crucial to minimize and control allograft rejection. Biopsy - the gold standard - is used only as a last resort due to its invasiveness, high cost, adverse events (e.g., bleeding, infection, etc.), and the time for reporting. To overcome these limitations, a renal computer-assisted diagnostic (Renal-CAD) system was developed to assess kidney transplant function. METHODS: The developed Renal-CAD system integrates data collected from two image-based sources and two clinical-based sources to assess renal transplant function. The imaging sources were the apparent diffusion coefficients (ADCs) extracted from 47 diffusion-weighted magnetic resonance imaging (DW-MRI) scans at 11 different b-values (b0, b50, b100, ..., b1000 s/mm 2 ), and the transverse relaxation rate (R2*) extracted from 30 blood oxygen level-dependent MRI (BOLD-MRI) scans at 5 different echo times (TEs = 2, 7, 12, 17, and 22 ms). Serum creatinine (SCr) and creatinine clearance (CrCl) were the clinical sources for kidney function evaluation. The Renal-CAD system initially performed kidney segmentation using the level-set method, followed by estimation of the ADCs from DW-MRIs and the R2* from BOLD-MRIs. ADCs and R2* estimates from 30 subjects that have both types of scans were integrated with their associated SCr and CrCl. The integrated biomarkers were then used as our discriminatory features to train and test a deep learning-based classifier, namely stacked autoencoders (SAEs) to differentiate non-rejection (NR) from acute rejection (AR) renal transplants. RESULTS: Using a leave-one-subject-out cross-validation approach along with SAEs, the Renal-CAD system demonstrated 93.3% accuracy, 90.0% sensitivity, and 95.0% specificity in differentiating AR from NR. Robustness of the Renal-CAD system was also confirmed by the area under the curve value of 0.92. Using a stratified tenfold cross-validation approach, the Renal-CAD system demonstrated its reproducibility and robustness by a diagnostic accuracy of 86.7%, sensitivity of 80.0%, specificity of 90.0%, and AUC of 0.88. CONCLUSION: The obtained results demonstrate the feasibility and efficacy of accurate, noninvasive identification of AR at an early stage using the Renal-CAD system.

A Generalized Deep Learning-Based Diagnostic System for Early Diagnosis of Various Types of Pulmonary Nodules.

A novel framework for the classification of lung nodules using computed tomography scans is proposed in this article. To get an accurate diagnosis of the detected lung nodules, the proposed framework integrates the following 2 groups of features: (1) appearance features modeled using the higher order Markov Gibbs random field model that has the ability to describe the spatial inhomogeneities inside the lung nodule and (2) geometric features that describe the shape geometry of the lung nodules. The novelty of this article is to accurately model the appearance of the detected lung nodules using a new developed seventh-order Markov Gibbs random field model that has the ability to model the existing spatial inhomogeneities for both small and large detected lung nodules, in addition to the integration with the extracted geometric features. Finally, a deep autoencoder classifier is fed by the above 2 feature groups to distinguish between the malignant and benign nodules. To evaluate the proposed framework, we used the publicly available data from the Lung Image Database Consortium. We used a total of 727 nodules that were collected from 467 patients. The proposed system demonstrates the promise to be a valuable tool for the detection of lung cancer evidenced by achieving a nodule classification accuracy of 91.20%.

Mobile Application for Ulcer Detection.

PURPOSE: The number of patients who are suffering from diabetes nowadays is increasing significantly. In some countries, the percentage of population who suffer from this disease can reach up to 20%. Diabetic patients have to deal with their medical conditions and any further complications that this disease may cause. One of the most common conditions is the Diabetic Foot Ulcer (DFU). The early detection of these ulcers can help and may save the life of diabetic patients. METHODS: This work proposes a mobile application for the detection of possible ulcers using a smart phone along with a mobile thermal camera (FLIR ONE). The proposed system captures thermal images of the feet from the thermal camera. The app that identifies ulcers was built using Android studio. The images were acquired to the Samsung S6 smart phone using the FLIR ONE SDK. Image processing techniques were deployed based on Open CV Library. The procedure of detecting possible ulcers was implemented based on analyzing the thermal distribution on the two feet. The developed application compares the difference between the temperature distribution on the two feet and checks if there is a Mean Temperature Difference (MTD) greater than 2.2oC (the value which indicates a possible ulcer development). RESULTS: The system was tested under simulated conditions by heating different locations of the subjects' feet to different temperature ranges; one image with temperature less than 2.2oC and another three images with temperature greater than 2.2oC. The system has successfully identified possible ulcer regions along with an image showing the location of the possible ulcers. CONCLUSIONS: This work is a very first step in developing a complete mobile thermal imaging system that can be validated clinically in the future.

Design of a stabilisation platform for Parkinson's disease patient.

There are an increased number of patients all over the world suffering from postural tremors and rest tremors, the types of tremors associated with Parkinson's disease and other neurodegenerative diseases such as amyotrophic lateral sclerosis. Currently, there is no cure for such disease and patients have to deal with their condition to continue their life normally with the existing of some helpful instruments. This work presents a self-stabilising Parkinson's disease (PD) Tray (platform) that can help them to carry objects that they hold with their hand. The proposed design includes a mechanical platform and en electronic system to control the tray and inhibits any vibrations of the base plate of the tray. An algorithm was developed that would take positional data from an Inertia movement sensors IMU, compute angles in degrees from its Euler angle raw data and then use those angles to control three servo motors in a direction counter to the changes in the IMU's position. The platform, was capable of stabilising the base of the tray such that objects placed in it would not be dropped. The tray was tested on simulating conditions and the result should that the mean absolute value of the acceleration values in X and Y directions were reduced from 2.23 m/sec2 to 0.26 m/sec2 in the X direction and from 1.41 m/sec2 to 0.34 m/sec2 in the Y direction.

Diabetic foot ulcer mobile detection system using smart phone thermal camera: a feasibility study.

BACKGROUND: Nowadays, the whole world is being concerned with a major health problem, which is diabetes. A very common symptom of diabetes is the diabetic foot ulcer (DFU). The early detection of such foot complications can protect diabetic patients from any dangerous stages that develop later and may require foot amputation. This work aims at building a mobile thermal imaging system that can be used as an indicator for possible developing ulcers. METHODS: The proposed system consists of a thermal camera connected to a Samsung smart phone, which is used to acquire thermal images. This thermal imaging system has a simulated temperature gradient of more than 2.2 °C, which represents the temperature difference (in the literature) than can indicate a possible development of ulcers. The acquired images are processed and segmented using basic image processing techniques. The analysis and interpretation is conducted using two techniques: Otsu thresholding technique and Point-to-Point mean difference technique. RESULTS: The proposed system was implemented under MATLAB Mobile platform and thermal images were analyzed and interpreted. Four testing images (feet images) were used to test this procedure; one image with any temperature variation to the feet, and three images with skin temperature increased to more than 2.2 °C introduced at different locations. With the two techniques applied during the analysis and interpretation stage, the system was successful in identifying the location of the temperature increase. CONCLUSION: This work successfully implemented a mobile thermal imaging system that includes an automated method to identify possible ulcers in diabetic patients. This may give diabetic patients the ability for a frequent self-check of possible ulcers. Although this work was implemented in simulated conditions, it provides the necessary feasibility to be further developed and tested in a clinical environment.

Parkinson's disease hand tremor detection system for mobile application.

Parkinson's disease currently affects millions of people worldwide and is steadily increasing. Many symptoms are associated with this disease, including rest tremor, bradykinesia, stiffness or rigidity of the extremities and postural instability. No cure is currently available for Parkinson's disease patients; instead most medications are for treatment of symptoms. This treatment depends on the quantification of these symptoms such as hand tremor. This work proposes a new system for mobile phone applications. The system is based on measuring the acceleration from the Parkinson's disease patient's hand using a mobile cell phone accelerometer. Recordings from 21 Parkinson's disease patients and 21 healthy subjects were used. These recordings were analysed using a two level wavelet packet analysis and features were extracted forming a feature vector of 12 elements. The features extracted from the 42 subjects were classified using a neural networks classifier. The results obtained showed an accuracy of 95% and a Kappa coefficient of 90%. These results indicate that a cell phone accelerometer can accurately detect and record rest tremor in Parkinson's disease patients.

Automated sleep stage identification system based on time-frequency analysis of a single EEG channel and random forest classifier.

In this work, an efficient automated new approach for sleep stage identification based on the new standard of the American academy of sleep medicine (AASM) is presented. The propose approach employs time-frequency analysis and entropy measures for feature extraction from a single electroencephalograph (EEG) channel. Three time-frequency techniques were deployed for the analysis of the EEG signal: Choi-Williams distribution (CWD), continuous wavelet transform (CWT), and Hilbert-Huang Transform (HHT). Polysomnographic recordings from sixteen subjects were used in this study and features were extracted from the time-frequency representation of the EEG signal using Renyi's entropy. The classification of the extracted features was done using random forest classifier. The performance of the new approach was tested by evaluating the accuracy and the kappa coefficient for the three time-frequency distributions: CWD, CWT, and HHT. The CWT time-frequency distribution outperformed the other two distributions and showed excellent performance with an accuracy of 0.83 and a kappa coefficient of 0.76.

Voiceless Arabic vowels recognition using facial EMG.

This work attempts to recognize the Arabic vowels based on facial electromyograph (EMG) signals, to be used for people with speech impairment and for human computer interface. Vowels were selected since they are the most difficult letters to recognize by people in Arabic language. Twenty subjects (7 females and 13 males) were asked to pronounce three Arabic vowels continuously in a random order. Facial EMG signals were recorded over three channels from the three main facial muscles that are responsible for speech. The EMG signals are then pre-processed to eliminate noise and interference signals. Segmentation procedure was implemented to extract the time event that corresponds to each vowel based on a moving standard deviation window. The accuracy of the segmentation procedure was found to be 94%. The recognition of the vowels was carried out by extracting features from the EMG in three domains: the temporal, the spectral, and the time frequency using the wavelet packet transform. Classification of the extracted features was then finally performed using different classification methods implemented in the WEKA software. The random forest classifier with time frequency features showed the best performance with an accuracy of 77% evaluated using a 10-fold cross-validation.

New automated detection method of OSA based on artificial neural networks using P-wave shape and time changes.

This paper describes a new method for automatic detection of obstructive sleep apnea (OSA) based on artificial neural networks (ANN) using regular electrocardiogram (ECG) recordings. ECG signals were pre-processed and segmented to extract the P-waves; then three P-wave features were extracted: the P-wave duration (T ( p )), the P-wave dispersion (P ( d )), and the time interval from the peak of the P-wave to the R-wave (T ( pr )). Combinations of the three features were used as features for classification using ANN. For each feature combination studied, 70% of the input data was used for training the ANN, 15% for validating, and 15% for testing the results. Perfect agreement between expert's scores and the ANN scores was achieved when the ANN was applied on T ( p ), P ( d ), and T ( pr ) taken together, while substantial agreements were achieved when applying the ANN on the feature combinations T ( p ) and P ( d ), and T ( p ) and T ( pr ).

Time frequency analysis for automated sleep stage identification in fullterm and preterm neonates.

This work presents a new methodology for automated sleep stage identification in neonates based on the time frequency distribution of single electroencephalogram (EEG) recording and artificial neural networks (ANN). Wigner-Ville distribution (WVD), Hilbert-Hough spectrum (HHS) and continuous wavelet transform (CWT) time frequency distributions were used to represent the EEG signal from which features were extracted using time frequency entropy. The classification of features was done using feed forward back-propagation ANN. The system was trained and tested using data taken from neonates of post-conceptual age of 40 weeks for both preterm (14 recordings) and fullterm (15 recordings). The identification of sleep stages was successfully implemented and the classification based on the WVD outperformed the approaches based on CWT and HHS. The accuracy and kappa coefficient were found to be 0.84 and 0.65 respectively for the fullterm neonates' recordings and 0.74 and 0.50 respectively for preterm neonates' recordings.