Biomedical Signal Processing / Medical Signal Processing / Biosignal Processing
Hamid Shafaatfar; Mehdi Taghizadeh; Morteza Valizadeh; Mohamad Hossein Fatehi
Volume 16, Issue 2 , September 2022, , Pages 147-158
Abstract
Automatic detection of cardiac arrhythmias is very important for the successful treatment of heart disease and machine learning is used for this purpose. To correctly classify arrhythmic classes, it is important to extract the appropriate features to distinguish between different classes. In this paper, ...
Read More
Automatic detection of cardiac arrhythmias is very important for the successful treatment of heart disease and machine learning is used for this purpose. To correctly classify arrhythmic classes, it is important to extract the appropriate features to distinguish between different classes. In this paper, a deep convolutional neural network is used to extract the feature. Due to the fact that the heart rates of different patients are very different, arrhythmia classes will have many intra-class changes. To reduce intra-class changes, each patient’s heart rate is mapped with a dedicated function to increase its resemblance to the heart rate of one of the training patient data’s. The proposed specific mapping reduces intra-class changes and significantly increases the classification accuracy of cardiac arrhythmias. To prove the effectiveness of the proposed method, its results were compared with several new studies based on three criteria for accuracy, sensitivity and specificity and on the same data set. The accuracy obtained is about 96.24%, which shows the better performance of the proposed method compared to other works.
Biomimetics
Hiwa Sufikarimi; Karim Mohammadi
Volume 11, Issue 4 , February 2018, , Pages 337-349
Abstract
In this paper, we tried to present a robust and reliable approach to object recognition by inspiring human visual system. A famous model, inspiring mammalian visual system, is HMAX (Hierarchical Model and X). It shows significant accuracy rates on object recognition tasks. However, there are some differences ...
Read More
In this paper, we tried to present a robust and reliable approach to object recognition by inspiring human visual system. A famous model, inspiring mammalian visual system, is HMAX (Hierarchical Model and X). It shows significant accuracy rates on object recognition tasks. However, there are some differences between this model and human visual system. Indeed cortex's functions are not properly modeled. Unrepeatability under fixed conditions, redundancy, high computing load and being slow are some drawbacks of HMAX. By modeling the secondary visual cortex and adding to the HMAX, we tried to introduce a more accurate model of the human visual system and cover the drawbacks of the previous models. The proposed approach functionally mimics the secondary visual cortex. Attending to high-level features, selecting discriminative and repeatable features, it has higher performance than standard HMAX. The added parts have negligible computation load. Therefore, it does not slow down this model. On the contrary, by selecting brief and useful features, the speed of the model is increased. The proposed approach is compared to the standard HMAX in terms of speed and accuracy rate. The results showed the advantage of proposed approach rather than the standard HMAX. In addition, the effect of the number of features and training images on their performance was shown. It is shown that the proposed approach has a better performance than the standard HMAX especially when the number of feature and training images is small.
Biomedical Signal Processing / Medical Signal Processing / Biosignal Processing
Masoumeh Rahimi; Mohammad Hasan Moradi; Farnaz Ghassemi
Volume 10, Issue 1 , May 2016, , Pages 59-68
Abstract
The aim of this paper is to study brain effective connectivity based on directed transform function (DTF) using granger causality method. This connectivity was calculated for recorded data in different states of attention and consciousness, forming four different classes: attention-consciousness, attention-unconsciousness, ...
Read More
The aim of this paper is to study brain effective connectivity based on directed transform function (DTF) using granger causality method. This connectivity was calculated for recorded data in different states of attention and consciousness, forming four different classes: attention-consciousness, attention-unconsciousness, inattention-consciousness, and inattention-unconsciousness. Some common indices were extracted and calculated from the connectivity matrices. Indices of these four classes were compared to see whether there is a significant difference among them or not. The Multivariate Autoregressive (MVAR) model was used to obtain the linear causal relations between channels. Furthermore, signals were divided into four frequency bands for more accurate investigation, and the existence of significant difference was investigated with two-way repeated measures test. Results indicated that and among twelve indices could show a significant difference (p<0.05) in five states out of six possible states. The only state that no feature was able to show a meaningful difference was inattention-consciousness, and inattention-unconsciousness.
Speech processing
Yaser Shekofteh; Farshad Almasganj
Volume 6, Issue 1 , June 2012, , Pages 17-33
Abstract
Recent researches show that nonlinear and chaotic behavior of the speech signal can be studied in the reconstructed phase space (RPS). Delay embedding theorem is a useful tool to study embedded speech trajectories in the RPS. Characteristics of the speech trajectories have rarely used in the practical ...
Read More
Recent researches show that nonlinear and chaotic behavior of the speech signal can be studied in the reconstructed phase space (RPS). Delay embedding theorem is a useful tool to study embedded speech trajectories in the RPS. Characteristics of the speech trajectories have rarely used in the practical speech recognition systems. Therefore, in this paper, a new feature extraction (FE) method is proposed based on parameters of vector AR (VAR) analysis over the speech trajectories. In this method, using filter and reflection matrices obtained from applying VAR analysis on static and dynamic information of the speech trajectory in the RPS, a high-dimensional feature vector can be achieved. Then, different transformation methods are utilized to attain final feature vectors with appropriate dimension. Results of discrete and continuous phoneme recognition over FARSDAT speech corpus show that the efficiency of the proposed FE method is better than other time-domain-based FE methods such as LPC and LPREF.
Biomedical Signal Processing / Medical Signal Processing / Biosignal Processing
Faride Ebrahimi; Mohammad Mikaili
Volume 4, Issue 2 , June 2010, , Pages 97-108
Abstract
Different biological signals including EEG, EOG, and EMG are recorded in sleep labs to diagnose sleep disorders. Data recorded during sleep is usually analyzed by sleep specialists visually. Since the sleep data is usually recorded for a long time period- namely a whole night- its visual inspection and ...
Read More
Different biological signals including EEG, EOG, and EMG are recorded in sleep labs to diagnose sleep disorders. Data recorded during sleep is usually analyzed by sleep specialists visually. Since the sleep data is usually recorded for a long time period- namely a whole night- its visual inspection and classification is a very demanding and time consuming task so automatic analysis can definitely facilitate that. The key to automatic sleep staging is to extract suitable features. In the current study two classes of features are extracted from EEG signal. The first group is the features calculated from the coefficients of wavelet packet transformation (WPT) and the second group consists of a number of frequency features and a time feature, the amplitude of EEG signal itself. These two sets of features were separately mapped on a two dimensional space by SOM neural networks. The mappings indicated that these features are highly discriminative in separating sleep stages automatically. The data extracted from awake and deep sleep EEGs were mapped on two totally different regions. The mapping also indicated that EEG signal is not enough to separate stages thoroughly, as extracted data from EEG during REM and the first stage of NREM are mapped on the same region. Data extracted from EEG signals in the second stage overlapped with other stages which are in agreement with physiological definition of sleep stages.
Zahra Amini; Vahid Abootalebi; Mohammad Taghi Sadeghi
Volume 4, Issue 4 , June 2010, , Pages 293-306
Abstract
The aim of this paper is to design a pattern recognition based system to detect P300 component in multi-channel electroencephalogram (EEG) trials. This system has two main blocks, feature extraction and classification. In feature extraction block, in addition to conventional features namely morphological, ...
Read More
The aim of this paper is to design a pattern recognition based system to detect P300 component in multi-channel electroencephalogram (EEG) trials. This system has two main blocks, feature extraction and classification. In feature extraction block, in addition to conventional features namely morphological, frequency and wavelet features, some new features included intelligent segmentation, common spatial pattern (CSP) and combined features (CSP + Segmentation) have also been used. Three criteria were used for evaluation and selection of a feature set by choosing a subset of the original features that contains most of essential information. Firstly, a statistical analysis has been applied for evaluating the fitness of each feature in discriminating between target and non target signals. Secondly, each of these six groups of features was evaluated by a Linear Discriminant Analysis (LDA) classifier. Furthermore by using Stepwise Linear Discriminant Analysis (SWLDA), the best set of features was selected. Among these six feature vectors, intelligent segmentation was seen to be most efficient in classification of these signals. In classification phase, two linear classifiers -LDA and SWLDA- were used. The algorithm was described here has tested with dataset II from the BCI competition 2005. In this research, the best result for P300 detection is 97.05% .This result have proven to be more accurate than the results of previous works carried out in this filed.
Zohre Dehghani Bidgoli; Mohammad Hossein Miranbaygi; Rasoul Malekfar; Ehsanollah Kabir; Tahere Khamechian
Volume 4, Issue 4 , June 2010, , Pages 307-316
Abstract
In this research, we investigated cancerous tissues from several organs of the human body using Raman spectroscopy. Different specimens with different pathologic labels (normal & cancerous) were borrowed from a pathology laboratory, and were investigated using two different Raman spectroscopy systems. ...
Read More
In this research, we investigated cancerous tissues from several organs of the human body using Raman spectroscopy. Different specimens with different pathologic labels (normal & cancerous) were borrowed from a pathology laboratory, and were investigated using two different Raman spectroscopy systems. Since one of the goals of this investigation was detection of cancer, independent of type of the system, we introduced some algorithms for removing systemic differences from the spectra. Then we removed noise and fluorescence signals using a new wavelet created with LWT. The best classification result was 83% in differentiating between normal and cancerous specimens using the SVM classifier
Biomedical Image Processing / Medical Image Processing
Meysam Torabi; Emadoddin Fatemizadeh
Volume 3, Issue 3 , June 2009, , Pages 213-225
Abstract
In this paper, an MRI-based diagnosing approach has been proposed which simultaneously analyzes T1-MR and T2-MR images. The dataset contains 120 cross-sectional images of abnormal and also normal brains as control group. Due to inherent proprieties of T1 and T2 images and their principal differences, ...
Read More
In this paper, an MRI-based diagnosing approach has been proposed which simultaneously analyzes T1-MR and T2-MR images. The dataset contains 120 cross-sectional images of abnormal and also normal brains as control group. Due to inherent proprieties of T1 and T2 images and their principal differences, particular features have been extracted from each image. Then, more meaningful data has been structured by automatically eliminating redundant data and generating a semi-linear combination of the remaining features. Considering the fact that Alzheimer's disease mainly damages the gray and white matter of the brain and knowing that these parts of the brain can be more clearly observed in T1 images, the classifier which works under a nonlinear structure, allocates more weight for processing the T1 images comparing to T2 image. The images, after being registered, have been processed in two groups of training and test sets. According to the results, three forth of the dataset which was obtained from Harvard University's dataset (The Whole Brain Atlas) has been correctly diagnosed.