Improving a cortical pyramidal neuron model's classification performance on a real-world ecg dataset by extending inputs.

Pyramidal neurons display a variety of active conductivities and complex morphologies that support nonlinear dendritic computation. Given growing interest in understanding the ability of pyramidal neurons to classify real-world data, in our study we applied both a detailed pyramidal neuron model and the perceptron learning algorithm to classify real-world ECG data. We used Gray coding to generate spike patterns from ECG signals as well as investigated the classification performance of the pyramidal neuron's subcellular regions. Compared with the equivalent single-layer perceptron, the pyramidal neuron performed poorly due to a weight constraint. A proposed mirroring approach for inputs, however, significantly boosted the classification performance of the neuron. We thus conclude that pyramidal neurons can classify real-world data and that the mirroring approach affects performance in a way similar to non-constrained learning.

A novel signal to image transformation and feature level fusion for multimodal emotion recognition.

Emotion is one of the most complex and difficult expression to be predicted. Nowadays, many recognition systems that use classification methods have focused on different types of emotion recognition problems. In this paper, we aimed to propose a multimodal fusion method between electroencephalography (EEG) and electrooculography (EOG) signals for emotion recognition. Therefore, before the feature extraction stage, we applied different angle-amplitude transformations to EEG-EOG signals. These transformations take arbitrary time domain signals and convert them two-dimensional images named as Angle-Amplitude Graph (AAG). Then, we extracted image-based features using a scale invariant feature transform method, fused these features originates basically from EEG-EOG and lastly classified with support vector machines. To verify the validity of these proposed methods, we performed experiments on the multimodal DEAP dataset which is a benchmark dataset widely used for emotion analysis with physiological signals. In the experiments, we applied the proposed emotion recognition procedures on the arousal-valence dimensions. We achieved (91.53%) accuracy for the arousal space and (90.31%) for the valence space after fusion. Experimental results showed that the combination of AAG image features belonging to EEG-EOG signals in the baseline angle amplitude transformation approaches enhanced the classification performance on the DEAP dataset.

Diversity in a signal-to-image transformation approach for EEG-based motor imagery task classification.

Nowadays, motor imagery-based brain-computer interfaces (BCIs) have been developed rapidly. In these systems, electroencephalogram (EEG) signals are recorded when a subject is involved in the imagination of doing any motor imagery movement like the imagination of the right/left hands, etc. In this paper, we sought to validate and enhance our previously proposed angle-amplitude transformation (AAT) technique, which is a simple signal-to-image transformation approach for the classification of EEG and MEG signals. For this purpose, we diversified our previous method and proposed four new angle-amplitude graph (AAG) representation methods for AAT transformation. These modifications were made on some points such as using different left/right side changing points at a different distance. To confirm the validity of the proposed methods, we performed experiments on the BCI Competition III Dataset IIIa, which is a benchmark dataset widely used for EEG-based multi-class motor imagery tasks. The procedure of proposed methods can be summarized in a concise manner as follows: (i) convert EEG signals to AAG images by using the proposed AAT transformation approaches; (ii) extract image features by employing Scale Invariant Feature Transform (SIFT)-based Bag of Visual Word (BoW); and (iii) classify features with k-Nearest Neighbor (k NN) algorithm. Experimental results showed that the changes in the baseline AAT approaches enhanced the classification performance on Dataset IIIa with an accuracy of 96.50% for two-class problem (left/right hand movement imaginations) and 97.99% for four-class problem (left/right hand, foot and tongue movement imaginations). These achievements are mainly due to the help of effective enhancements on AAG image representations. Graphical Abstract The flow diagram of the proposed methodology.

Assessment of CSP-based two-stage channel selection approach and local transformation-based feature extraction for classification of motor imagery/movement EEG data.

The main idea of brain-computer interfaces (BCIs) is to facilitate the lives of patients having difficulties to move their muscles due to a disorder of their motor nervous systems but healthy cognitive functions. BCIs are usually electroencephalography (EEG)-based, and the success of the BCIs relies on the precision of signal preprocessing, detection of distinctive features, usage of suitable classifiers and selection of effective channels. In this study, a two-stage channel selection and local transformation-based feature extraction are proposed for the classification of motor imagery/movement tasks. In the first stage of the channel selection, the channels were combined according to the neurophysiological information about brain functions acquired from the literature, then averaged and a single channel was formed. In the second stage, selective channels were specified with the common spatial pattern-linear discriminant analysis (CSP-LDA)-based sequential channel removal. After the channel selection phase, the feature extraction was carried out with local transformation-based methods (LTBM): local centroid pattern (LCP), one-dimensional-local gradient pattern (1D-LGP), local neighborhood descriptive pattern (LNDP) and one-dimensional-local ternary pattern (1D-LTP). The distinctions and deficiencies of these methods were compared with other methods in the literature and the classification performances of the k-nearest neighbor (k-NN) and the support vector machines (SVM) were evaluated. As a result, the proposed methods yielded the highest average classification accuracies as 99.34%, 95.95%, 98.66% and 99.90% with the LCP, 1D-LGP, LNDP and 1D-LTP when using k-NN, respectively. The two-stage channel selection and 1D-LTP method showed promising results for recognition of motor tasks. The LTBM will contribute to the development of EEG-based BCIs with the advantages of high classification accuracy, easy implementation and low computational complexity.

A Quasi-probabilistic distribution model for EEG Signal classification by using 2-D signal representation.

BACKGROUND AND OBJECTIVE: Electroencephalography (EEG) is a method that measures and records the electrical activity of the human brain. These biomedical signals are currently being actively used in many research fields and have a wide range of potential uses in brain-computer interfaces (BCIs). The main aim of the present work is to improve the classification of EEG patterns for EEG-based BCI systems. METHODS: In this paper, we presented a classification approach for EEG-based BCIs. For this purpose, in the training stage, 2-D representations of signals were extracted and a quasi-probabilistic learning model was built for binary classification. In the testing stage, the estimation of class membership probability was performed with an untrained sub-data set. To confirm the validity of the proposed method, we conducted experiments on the BCI Competition 2003 Data Sets (Ia and Ib). The classification performances were evaluated for accuracy, sensitivity, specificity and F-measure measurements using the five-fold leave-one-out cross-validation technique ten times. RESULTS: The proposed method yielded an average classification accuracy of 95.54% (with sensitivity and specificity of 100.00% and 91.80% respectively) for Data Set Ia and accuracy of 72.37% (with sensitivity and specificity of 75.76% and 69.77% respectively) for Data Set Ib, which are the highest rates ever reported for both data sets. CONCLUSIONS: It is apparent from the results that the proposed method has potential and can assist in the development of effective EEG-based BCIs. The advantage of this method lies in its relatively simple algorithm and easy computational implementation. The experimental results also showed that the selection of relevant channels is an important step in developing efficient EEG-based BCI systems.

Blood Vessel Extraction in Color Retinal Fundus Images with Enhancement Filtering and Unsupervised Classification.

Retinal blood vessels have a significant role in the diagnosis and treatment of various retinal diseases such as diabetic retinopathy, glaucoma, arteriosclerosis, and hypertension. For this reason, retinal vasculature extraction is important in order to help specialists for the diagnosis and treatment of systematic diseases. In this paper, a novel approach is developed to extract retinal blood vessel network. Our method comprises four stages: (1) preprocessing stage in order to prepare dataset for segmentation; (2) an enhancement procedure including Gabor, Frangi, and Gauss filters obtained separately before a top-hat transform; (3) a hard and soft clustering stage which includes K-means and Fuzzy C-means (FCM) in order to get binary vessel map; and (4) a postprocessing step which removes falsely segmented isolated regions. The method is tested on color retinal images obtained from STARE and DRIVE databases which are available online. As a result, Gabor filter followed by K-means clustering method achieves 95.94% and 95.71% of accuracy for STARE and DRIVE databases, respectively, which are acceptable for diagnosis systems.

SNORAP: A Device for the Correction of Impaired Sleep Health by Using Tactile Stimulation for Individuals with Mild and Moderate Sleep Disordered Breathing.

Sleep physiology and sleep hygiene play significant roles in maintaining the daily lives of individuals given that sleep is an important physiological need to protect the functions of the human brain. Sleep disordered breathing (SDB) is an important disease that disturbs this need. Snoring and Obstructive Sleep Apnea Syndrome (OSAS) are clinical conditions that affect all body organs and systems that intermittently, repeatedly, with at least 10 s or more breathing stops that decrease throughout the night and disturb sleep integrity. The aim of this study was to produce a new device for the treatment of patients especially with position and rapid eye movement (REM)-dependent mild and moderate OSAS. For this purpose, the main components of the device (the microphone (snore sensor), the heart rate sensor, and the vibration motor, which we named SNORAP) were applied to five volunteer patients (male, mean age: 33.2, body mass index mean: 29.3). After receiving the sound in real time with the microphone, the snoring sound was detected by using the Audio Fingerprint method with a success rate of 98.9%. According to the results obtained, the severity and the number of the snoring of the patients using SNORAP were found to be significantly lower than in the experimental conditions in the apnea hypopnea index (AHI), apnea index, hypopnea index, in supine position's AHI, and REM position's AHI before using SNORAP (Paired Sample Test, p < 0.05). REM sleep duration and nocturnal oxygen saturation were significantly higher when compared to the group not using the SNORAP (Paired Sample Test, p < 0.05).

Automatic retrieval of shoeprint images using blocked sparse representation.

Shoe marks are regarded as remarkable clues which can be usually detected in crime scenes where forensic experts use them for investigating crimes and identifying the criminals. Hence, developing a robust method for matching shoeprints with one another is of critical significance which can be used for recognizing criminals. In this paper, a promising method is proposed for retrieving shoe marks by means of developing blocking sparse representation technique. In the proposed method, the queried image was divided into two blocks. Then, two sparse representations are extracted for each queried image through approximate ℓ1 minimizing method. Also, the referenced database is categorized into two parts and two separate dictionaries are developed via them. Next, using the blocks, the total errors of classes are measured by resetting the coefficients related to other classes into zero. The performance of the proposed method was evaluated via the following methods Wright's sparse representation, extracting shoeprint image local and global features by Fourier transform, extracting shoeprint image features by Gabor transform after the image is rotated and extracting the corners of shoeprint image by Hessian and Harris' multi-scale detectors and SIFT descriptors. Accurate detection score was obtained in terms of the ratio of the number of accurately detected images to the total test images. The results of simulations indicated that the proposed method was highly effective and efficient in retrieving shoe marks, whole shoeprints, partial toe and heel shoeprints. Furthermore, it was found that the proposed method had better performance than the other methods with which it was compared. Accurate identification rate according to cumulative match score for the first n matches was measured. That is to say, the proposed method accurately recognized 99.47% of whole shoeprints, 80.53% of partial toe shoeprints and 79.47% of partial heel shoeprints in the first rank. Also, the proposed method was compared with the other methods in terms of rotation and scale distortions. The results indicated that the proposed method was resistant to these distortions.

Identification of suitable fundus images using automated quality assessment methods.

Retinal image quality assessment (IQA) is a crucial process for automated retinal image analysis systems to obtain an accurate and successful diagnosis of retinal diseases. Consequently, the first step in a good retinal image analysis system is measuring the quality of the input image. We present an approach for finding medically suitable retinal images for retinal diagnosis. We used a three-class grading system that consists of good, bad, and outlier classes. We created a retinal image quality dataset with a total of 216 consecutive images called the Diabetic Retinopathy Image Database. We identified the suitable images within the good images for automatic retinal image analysis systems using a novel method. Subsequently, we evaluated our retinal image suitability approach using the Digital Retinal Images for Vessel Extraction and Standard Diabetic Retinopathy Database Calibration level 1 public datasets. The results were measured through the F1 metric, which is a harmonic mean of precision and recall metrics. The highest F1 scores of the IQA tests were 99.60%, 96.50%, and 85.00% for good, bad, and outlier classes, respectively. Additionally, the accuracy of our suitable image detection approach was 98.08%. Our approach can be integrated into any automatic retinal analysis system with sufficient performance scores.

Simple methods for segmentation and measurement of diabetic retinopathy lesions in retinal fundus images.

Diabetic retinopathy (DR) is one of the most important complications of diabetes mellitus, which causes serious damages in the retina, consequently visual loss and sometimes blindness if necessary medical treatment is not applied on time. One of the difficulties in this illness is that the patient with diabetes mellitus requires a continuous screening for early detection. So far, numerous methods have been proposed by researchers to automate the detection process of DR in retinal fundus images. In this paper, we developed an alternative simple approach to detect DR. This method was built on the inverse segmentation method, which we suggested before to detect Age Related Macular Degeneration (ARMDs). Background image approach along with inverse segmentation is employed to measure and follow up the degenerations in retinal fundus images. Direct segmentation techniques generate unsatisfactory results in some cases. This is because of the fact that the texture of unhealthy areas such as DR is not homogenous. The inverse method is proposed to exploit the homogeneity of healthy areas rather than dealing with varying structure of unhealthy areas for segmenting bright lesions (hard exudates and cotton wool spots). On the other hand, the background image, dividing the retinal image into high and low intensity areas, is exploited in segmentation of hard exudates and cotton wool spots, and microaneurysms (MAs) and hemorrhages (HEMs), separately. Therefore, a complete segmentation system is developed for segmenting DR, including hard exudates, cotton wool spots, MAs, and HEMs. This application is able to measure total changes across the whole retinal image. Hence, retinal images that belong to the same patients are examined in order to monitor the trend of the illness. To make a comparison with other methods, a Naïve Bayes method is applied for segmentation of DR. The performance of the system, tested on different data sets including various qualities of retinal fundus images, is over 95% in detection of the optic disc (OD), and 90% in segmentation of the DR.

A statistical segmentation method for measuring age-related macular degeneration in retinal fundus images.

Day by day, huge amount of information is collected in medical databases. These databases include quite interesting information that could be exploited in diagnosis of illnesses and medical treatment of patients. Classification of these data is getting harder as the databases are expanded. On the other hand, automated image analysis and processing is one of the most promising areas of computer vision used in medical diagnosis and treatment. In this context, retinal fundus images, offering very high resolutions that are sufficient for most of the clinical cases, provide many indications that could be exploited in diagnosing and screening retinal degenerations or diseases. Consequently, there is a strong demand in developing automated evaluation systems to utilize the information stored in the medical databases. This study proposes an automatic method for segmentation of ARMD in retinal fundus images. The method used in the automated system extracts lesions of the ARMD by employing a statistical method. In order to do this, the statistical segmentation method is first used to extract the healthy area of the macula that is more familiar and regular than the unhealthy parts. Here, characteristic images of the patterns of the macula are extracted and used to segment the healthy textures of an eye. In addition to this, blood vessels are also extracted and then classified as healthy regions. Finally, the inverse image of the segmented image is generated which determines the unhealthy regions of the macula. The performance of the method is examined on various quality retinal fundus images. Segmented images are also compared with consecutive images of the same patient to follow up the changes in the disease.

Automatic segmentation of age-related macular degeneration in retinal fundus images.

Every year an increasing number of people are affected by age-related macular degeneration (ARMD). Consequently, vast amount of information is accumulated in medical databases and manual classification of this information is becoming more and more difficult. Therefore, there is an increasing interest in developing automated evaluation methods to follow up the diseases. In this paper, we have presented an automatic method for segmenting the ARMD in retinal fundus images. Previously used direct segmentation techniques, generating unsatisfactory results in some cases, are more complex and costly than our inverse method. This is because of the fact that the texture of unhealthy areas of macula is quite irregular and varies from eye to eye. Therefore, a simple inverse segmentation method is proposed to exploit the homogeneity of healthy areas of the macula rather than unhealthy areas. This method first extracts healthy areas of the macula by employing a simple region growing method. Then, blood vessels are also extracted and classified as healthy regions. In order to produce the final segmented image, the inverse image of the segmented image is generated as unhealthy region of the macula. The performance of the method is examined on various qualities of retinal fundus images. The segmentation method without any user involvement provides over 90% segmentation accuracy. Segmented images with reference invariants are also compared with consecutive images of the same patient to follow up the changes in the disease.