Ashok A. Ghatol

Artificial Neural Network based cardiac arrhythmia classification using ECG signal data

In this paper we proposed a automated Artificial Neural Network (ANN) based classification system for cardiac arrhythmia using standard 12 lead ECG recordings. In this study, we are mainly interested in producing high confident arrhythmia classification results to be applicable in diagnostic decision support systems. In arrhythmia analysis, it is unavoidable that some attribute values of a person would be missing. Therefore we have replaced these missing attributes by closest column value of the concern class. Multilayer percepron (MLP) feedforward neural network model with static backpropagation algorithm is used to classify arrhythmia cases into normal and abnormal classes. Networks models are trained and tested for UCI ECG arrhythmia data set. This data set is a good environment to test classifiers as it is incomplete and ambiguous bio-signal data collected from total 452 patient cases. The classification performance is evaluated using six measures; sensitivity, specificity, classification accuracy, mean squared error (MSE), receiver operating characteristics (ROC) and area under curve (AUC). Our experimental results give 86.67% testing classification accuracy.

Feature selection for medical diagnosis: Evaluation for cardiovascular diseases

Machine learning has emerged as an effective medical diagnostic support system. In a medical diagnosis problem, a set of features that are representative of all the variations of the disease are necessary. The objective of our work is to predict more accurately the presence of cardiovascular disease with reduced number of attributes. We investigate intelligent system to generate feature subset with improvement in diagnostic performance. Features ranked with distance measure are searched through forward inclusion, forward selection and backward elimination search techniques to find subset that gives improved classification result. We propose hybrid forward selection technique for cardiovascular disease diagnosis. Our experiment demonstrates that this approach finds smaller subsets and increases the accuracy of diagnosis compared to forward inclusion and back-elimination techniques.

ECG arrhythmia classification using modular neural network model

This research is on presenting a new approach for cardiac arrhythmia disease classification. The proposed method uses Modular neural network (MNN) model to classify arrhythmia into normal and abnormal classes. We have performed experiments on UCI Arrhythmia data set. Missing attribute values of this data set are replaced by closest column value of the concern class. We have constructed neural network model by varying number of hidden layers from one to three and are trained by varying training percentage in data set partitions. In this study, we are mainly interested in producing high confident arrhythmia classification results to be applicable in diagnostic decision support systems. This data set is a good environment to test classifiers as it is incomplete and ambiguous bio-signal data collected from total 452 patient cases. The classification performance is evaluated using six measures; sensitivity, specificity, classification accuracy, mean squared error (MSE), receiver operating characteristics (ROC) and area under curve (AUC). The experimental results presented in this paper show that up to 82.22% testing classification accuracy can be obtained.

Artificial Neural Network Models based Cardiac Arrhythmia Disease Diagnosis from ECG Signal Data

Changes in the normal rhythm of a human heart may result in different cardiac arrhythmias, which may be immediately causes irreparable damage to the heart sustained over long periods of time. The ability to automatically identify arrhythmias from ECG recordings is important for clinical diagnosis and treatment. In this paper we proposed an Artificial Neural Network (ANN) based cardiac arrhythmia disease diagnosis system using standard 12 lead ECG signal recordings data. In this study, we are mainly interested in classifying disease in normal and abnormal classes. We have used UCI ECG signal data to train and test three different ANN models. In arrhythmia analysis, it is unavoidable that some attribute values of a person would be missing. Therefore we have replaced these missing attributes by closest column value of the concern class. ANN models are trained by static backpropagation algorithm with momentum learning rule to diagnose cardiac arrhythmia. The classification performance is evaluated using measures such as mean squared error (MSE), classification specificity, sensitivity, accuracy, receiver operating characteristics (ROC) and area under curve (AUC). Out of three different ANN models Multilayer perceptron ANN model have given very attractive classification results in terms of classification accuracy and sensitivity of 86.67% and 93.75% respectively while Modular ANN have given 93.1% classification specificity. General Terms Machine Learning, Pattern Classification.

A robust and reliable ECG pattern classification using QRS morphological features and ANN

This paper describes electrocardiogram (ECG) pattern classification using QRS morphological features and the artificial neural network. Four types of ECG patterns were chosen from the MIT-BIH database to be classified, including normal sinus rhythm, premature ventricular contraction, atrial premature beat and left bundle branch block beat. Authors propose a set of six ECG morphological features to reduce the feature vector size considerably to make the training process faster, and realize a simple but effective ECG heartbeat extraction scheme. Three types of artificial neural network models, MLP, RBF neural networks and support vector machine were separately trained and tested for ECG pattern classification and the experimental results of the different models have been compared. The MLP network exhibited the best performance and reached an overall test accuracy of 99.65%, while, RBF and SVM network reached 99.1% and 99.5% respectively. The performance of these classifiers was also evaluated in presence of additive white Gaussian noise. MLP network was found to be more robust in this respect.

Modular neural network model based foetal state classification

Cardiotocography (CTG) is a simultaneous recording of foetal heart rate (FHR) and uterine contractions (UC) and it is one of the most common diagnostic techniques to evaluate maternal and foetal well-being during pregnancy and before delivery. Assessment of the foetal state can be verified only after delivery using the foetal (newborn) outcome data. One of the most important features defining the abnormal foetal outcome is low birth weight. This paper proposes a multi-class classification algorithm using Modular neural network (MNN) models. It tries to boost two conflicting main objectives of multi-class classifiers: a high correct classification rate level and a high classification rate for each class. Using a Cardiotocography database of normal, suspect and pathological cases, we trained MNN classifiers with 23 real valued diagnostic features collected from total 2126 foetal CTG signal recordings data from UCI Machine Learning Repository. We used the classification in a detection process. The proposed methodology is presented, which then is tested on UCI Cardiotocography unseen testing data sets. Experimental results are promising paving the way for further research in that direction.

Selection of Mother Wavelet for Image Compression on Basis of Image

Recently discrete wavelet transform and wavelet packet has emerged as popular techniques for image compression. This paper compares compression performance of Daubechies, Biorthogonal, Coiflets and other wavelets along with results for different frequency images. Based on the result, we propose that proper selection of mother wavelet on the basis of nature of images and improve the quality and compression ratio remarkably

Artificial Neural Network Based Cardiac Arrhythmia Disease Diagnosis

Changes in the normal rhythm of a human heart may result in different cardiac arrhythmias, which may be immediately fatal or cause irreparable damage to the heart sustained over long periods of time. The ability to automatically identify arrhythmias from ECG recordings is important for clinical diagnosis and treatment. In this paper we proposed an Artificial Neural Network (ANN) based cardiac arrhythmia disease diagnosis system using standard 12 lead ECG signal recordings data. In this study, we are mainly interested in classifying disease in normal and abnormal classes. We have used UCI ECG signal data to train and test three different ANN models. In arrhythmia analysis, it is unavoidable that some attribute values of a person would be missing. Therefore we have replaced these missing attributes by closest column value of the concern class. ANN models are trained by static backpropagation algorithm with momentum learning rule to diagnose cardiac arrhythmia. The classification performance is evaluated using measures such as mean squared error (MSE), classification specificity, sensitivity, accuracy, receiver operating characteristics (ROC) and area under curve (AUC). Out of three different ANN models Multilayer perceptron ANN model have given very attractive classification results in terms of classification accuracy and sensitivity of 86.67% and 93.75% respectively while Modular ANN have given 93.1% classification specificity.

Arrhythmia disease classification using Artificial Neural Network model

In this paper we proposed an automated Artificial Neural Network (ANN) based classification system for cardiac arrhythmia disease using standard 12 lead ECG signal recordings. In this study, we are mainly interested in classifying different arrhythmia types (classes) using multilayer peceptron (MLP) model. We have used UCI ECG signal data to train and test MLP network model. For this multi class classification we used one arrhythmia class against normal arrhythmia class. Different arrhythmia types include coronary artery disease, old anterior myocardial infarction, old inferior myocardial infarction, sinus tachycardia, sinus bradycardia, right bundle branch block etc. In arrhythmia analysis, it is unavoidable that some attribute values of a person would be missing. Therefore we have replaced these missing attributes by closest column value of the concern class. MLP feedforward neural network model is trained by static backpropagation algorithm with momentum learning rule to classify cardiac arrhythmia classes. The classification performance is evaluated using measures such as classification accuracy, training, testing and cross validation mean squared error (MSE), percentage correct, receiver operating characteristics (ROC) and area under curve (AUC). From careful and exhaustive experimentation, we reached to the conclusion that proposed classifier gives best classification results in terms of classification accuracy of 100 % for classes 1 and 2, 98.72%, 97.4%, 94.25%, 92.1% for classes 4, 5, 2 and 10 respectively.

An effective feature set for ECG pattern classification

In this paper, QRS morphological features and the artificial neural network method was used for Electrocardiogram (ECG) pattern classification. Four types of ECG patterns were chosen from the MIT-BIH database to be recognized, including normal sinus rhythm, premature ventricular contraction, atrial premature beat and left bundle branch block beat. Authors propose a set of six ECG morphological features to reduce the feature vector size considerably and make the training process fast in addition to a simple but effective ECG heartbeat extraction scheme. Three types of artificial neural network models, MLP, RBF neural networks and SOFM were separately trained and tested for ECG pattern recognition and the experimental results of the different models have been compared. The MLP network exhibited the best performance and reached an overall test accuracy of 99.65%, and RBF and SOFM network both reached 99.1%. The performance of these classifiers was also evaluated in presence of additive Gaussian noise. MLP network was found to be more robust in this respect.