Sadık Kara

A system to diagnose atherosclerosis via wavelet transforms, principal component analysis and artificial neural networks

Abstract In this study, Doppler ultrasound signals were acquired from carotid arteries of 82 patients with atherosclerosis and 95 healthy volunteers. We have employed discrete wave transform (DWT) of Doppler signals and power spectral density graphics of these decomposed signals using Welch method. After that, we have performed Principal component analysis (PCA) for data reduction and ANN in order to distinguish between atherosclerosis and healthy subjects. After the training phase, testing of the artificial neural network (ANN) was established. The overall results show that 97.9% correct classification was achieved, whereas two false classifications have been observed for the test group of 97 people. In conclusion we are proposing a complimentary expert system that can be coupled to software of the ultrasonic Doppler devices. The diagnosis performances of this study show the advantages of this system: it is rapid, easy to operate, noninvasive, inexpensive and making a decision without hesitation.

Classification of electro-oculogram signals using artificial neural network

Abstract This research is concentrated on the diagnosis of subnormal eye through the analysis of Electrooculography (EOG) signals with the help of Artificial Neural Network (ANN). Multilayer feed forward ANN trained with a Levenberg Marquart (LM) backpropagation algorithm was implemented. The designed classification structure has about 94.1% sensitivity, 93.3% specifity and positive prediction is calculated to be 94.1%. The end results are classified as normal and subnormal eye. Testing results were found to be compliant with the expected results that are derived from the physicians direct diagnosis. The benefit of the system is to assist the physician to make the final decision without hesitation. With the future evolution of this system tested on a more populated subject groups, there is always potential for on-line implementation as an auxiliary diagnostic tool on the Electrophysiology machines.

Application of autoregressive and Fast Fourier Transform spectral analysis to tricuspid and mitral valve stenosis

Tricuspid and mitral valve flow area was determined from an apical four-chamber view. Doppler signals were recorded from normal subjects and patients with tricuspid and mitral valve stenosis by using a pulsed Doppler unit. The location of sample volume was chosen at the ventricular side of the valve orifice and within the right ventricular tract. This was done with the aid of an integrated cardiac imaging facility. The analog signal at the output of the Doppler unit was sampled and digitized using an analog/digital interface board and transferred to a personal computer. The data were then analyzed using the Fast Fourier Transform (FTT) and autoregressive (AR) modeling methods of spectral analysis and all the sonograms were obtained. Statistical comparison between the FFT and AR methods was made. The results show that the AR method offers a superior performance over the FFT method as regards the assessment of tricuspid and mitral valve stenosis.

Comparison of multilayer perceptron training algorithms for portal venous doppler signals in the cirrhosis disease

Abstract In this study, we developed an expert diagnostic system for the interpretation of the portal vein Doppler signals belong the patients with cirrhosis and healthy subjects using signal processing and Artificial Neural Network (ANN) methods. Power spectral densities (PSD) of these signals were obtained to input of ANN using Short Time Fourier Transform (STFT) method. The four layered Multilayer Perceptron (MLP) training algorithms that we have built had given very promising results in classifying the healthy and cirrhosis. For prediction purposes, it has been presented that Levenberg Marquardt training algorithm of MLP network employing backpropagation works reasonably well. The diagnosis performance of the study shows the advantages of this system: It is rapid, easy to operate, noninvasive and not expensive. This system is of the better clinical application over others, especially for earlier survey of population. The stated results show that the proposed method can make an effective interpretation and point out the ability of design of a new intelligent assistance diagnosis system.

Recognition of early phase of atherosclerosis using principles component analysis and artificial neural networks from carotid artery Doppler signals

Abstract Atherosclerosis means thickening and hardening of the arteries, which has dramatic effects on blood pressure, resistance and blood flow. Since angiography is invasive and has a relatively high cost, non-invasive ultrasonic Doppler sonography is generally recommended to diagnose of athersosclerosis. In this study, we have employed the sonograms depicted from Autoregressive (AR) modeling, Principles component analysis (PCA) for data reduction of Doppler sonograms and artificial neural networks (ANN) in order to distinguish between atherosclerosis and healthy subjects. The fuzzy appearance of the carotid artery Doppler signals makes physicians suspicious about the existence of diseases and causes false diagnosis. Our technique gets around this problem using ANN to decide and assist the physician to make the final judgment in confidence. The stated results show that training time and processing complexity were reduced using PCA-ANN architecture however the proposed method can make an effective interpretation and ANN classified Doppler signals successfully.

Fuzzy clustering complex-valued neural network to diagnose cirrhosis disease

In this study, fuzzy clustering complex-valued neural network (FCCVNN) was proposed to classify portal vein Doppler signals recorded from 54 patients with cirrhosis and 36 healthy subjects. This proposed neural network is a new model for biomedical pattern classification. The FCCVNN was composed of three phases: fuzzy clustering, calculation of FFT values and complex-valued neural network (CVNN). In first phase, fuzzy clustering was done to reduce the number of segments in training pattern. After that, FFT values of Doppler signals were calculated for pre-processing and then obtained values, which include real and imaginary components, were used as the inputs of the CVNN for classification of Doppler signals. Classification results of FCCVNN were evaluated by the different performance evaluation criterion in literature. It shows that Doppler signals were classified successfully with 100% correct classification rate using the proposed method. Moreover, the rates of sensitivity and specificity were calculated as 100% using FCCVNN method. These results were seen to be appropriate with the expected results that are derived from physicians direct diagnosis. This method would be assisted the physician to make the final decision.

Log Energy Entropy-Based EEG Classification with Multilayer Neural Networks in Seizure

In this study, normal EEG series recorded from healthy volunteers and epileptic EEG series recorded from patients within and without seizure are classified by using Multilayer Neural Network (MLNN) architectures with respect to several time domain entropy measures such as Shannon Entropy (ShanEn), Log Energy Entropy (LogEn), and Sample Entropy (Sampen). In tests, the MLNN is performed with several numbers of neurons for both one hidden layer and two hidden layers. The results show that segments in seizure have significantly lower entropy values than normal EEG series. This result indicates an important increase of EEG regularity in epilepsy patients. The LogEn approach, which has not been experienced in EEG classification yet, provides the most reliable features into the EEG classification with very low absolute error as 0.01. In particular, the MLNN can be proposed to distinguish the seizure activity from the seizure-free epileptic series where the LogEn values are considered as signal features that characterize the degree of EEG complexity. The highest classification accuracy is obtained for one hidden layer architecture.

Spectral analysis of photoplethysmographic signals: The importance of preprocessing

Abstract Heart rate variability (HRV) is an important and useful index to assess the responses of the autonomic nervous system (ANS). HRV analysis is performed using electrocardiography (ECG) or photoplethysmography (PPG) signals which are typically subject to noise and trends. Therefore, the elimination of these undesired conditions is very important to achieve reliable ANS activation results. The purpose of this study was to analyze and compare the effects of preprocessing on the spectral analysis of HRV signals obtained from PPG waveform. Preprocessing consists of two stages: filtering and detrending. The performance of linear Butterworth filter is compared with nonlinear weighted Myriad filter. After filtering, two different approaches, one based on least squares fitting and another on smoothness priors, were used to remove trends from the HRV signal. The results of two filtering and detrending methods were compared for spectral analysis accomplished using periodogram, Welchs periodogram and Burgs method. The performance of these methods is presented graphically and the importance of preprocessing clarified by comparing the results. Although both filters have almost the same performance in the results, the smoothness prior detrending approach was found more successful in removing trends that usually appear in the low frequency bands of PPG signals. In conclusion, the results showed that trends in PPG signals are altered during spectral analysis and must be removed prior to HRV analysis.

Diagnosis of atherosclerosis from carotid artery Doppler signals as a real-world medical application of artificial immune systems

In this study, we have employed the maximum envelope of the carotid artery Doppler sonograms derived from Fast Fourier Transformation-Welch Method and artificial immune systems in order to distinguish between atherosclerosis and healthy subjects. In this classification problem, the used artificial immune system has reached to 99.33% classification accuracy using 10-fold Cross Validation (CV) method with only two system units which reduced classification time considerably. This success shows that whereas artificial immune systems is a new research area, one can utilize from this new field to reach high performance for his problem.

Complex-valued wavelet artificial neural network for Doppler signals classifying

OBJECTIVE In this paper, the new complex-valued wavelet artificial neural network (CVWANN) was proposed for classifying Doppler signals recorded from patients and healthy volunteers. CVWANN was implemented on four different structures (CVWANN-1, -2, -3 and -4). MATERIALS AND METHODS In this study, carotid arterial Doppler ultrasound signals were acquired from left carotid arteries of 38 patients and 40 healthy volunteers. The patient group had an established diagnosis of the early phase of atherosclerosis through coronary or aortofemoropopliteal angiographies. In implemented structures in this paper, Haar wavelet and Mexican hat wavelet functions were used as real and imaginary parts of activation function on different sequence in hidden layer nodes. CVWANN-1, -2 -3 and -4 were implemented by using Haar-Haar, Mexican hat-Mexican hat, Haar-Mexican hat, Mexican hat-Haar as real-imaginary parts of activation function in hidden layer nodes, respectively. RESULTS AND CONCLUSION In contrast to CVWANN-2, which reached classification rates of 24.5%, CVWANN-1, -3 and -4 classified 40 healthy and 38 unhealthy subjects for both training and test phases with 100% correct classification rate using leave-one-out cross-validation. These networks have 100% sensitivity, 100% specifity and average detection rate is calculated as 100%. In addition, positive predictive value and negative predictive value were obtained as 100% for these networks. These results shown that CVWANN-1, -3 and -4 succeeded to classify Doppler signals. Moreover, training time and processing complexity were decreased considerable amount by using CVWANN-3. As conclusion, using of Mexican hat wavelet function in real and imaginary parts of hidden layer activation function (CVWANN-2) is not suitable for classifying healthy and unhealthy subjects with high accuracy rate. The cause of unsuitability (obtaining the poor results in CVWANN-2) is lack of harmony between type of activation function in hidden layer and type of input signals in neural network.