Classification of Electrocardiogram (ECG) plays an important role in clinical diagnosis of cardiac diseases. In this paper, we introduce an ECG beat classification system using convolutional neural networks (CNNs). The proposed model integrates two main parts, feature extraction and classification, of ECG pattern recognition system. This model automatically learns a suitable feature representation from raw ECG data and thus negates the need of hand-crafted features. By using a small and patient-specific training data, the proposed classification system efficiently classified ECG beats into five different classes recommended by Association for Advancement of Medical Instrumentation (AAMI). ECG signal from 44 recordings of the MIT-BIH database are used to evaluate the classification performance and the results demonstrate that the proposed approach achieves a significant classification accuracy and superior computational efficiency than most of the state-of-the-art methods for ECG signal classification.

An Automated ECG Beat Classification System Using Convolutional Neural Networks

This paper aims at developing a wearable system able to recognize the most significant cardiac arrhythmias through an efficient algorithm, in terms of low computational cost and memory usage, implementable in a portable, real-rime hardware. In addition, it must respect the specifications of good specificity and sensitivity, in order to permit a positive clinical validation. The hardware is constituted of a general propose microcontroller, which is able to acquire electro-cardiogram signal (ECG), perform analog to digital conversion and extract QRS complex. The algorithm classifies QRS complexes as normal or pathologic by means of selected features obtained from discrete fourier transform (DFT). Furthermore, a spatial wavelet pre-filter is also investigated to obtain an enhanced QRS complex discrimination. In particular, pattern recognition of QRS complex is performed from binding minimal architecture of neural network as Kohonen self organizing map (KSOM). Experimental results were validated by means of MIT-BIH arrhythmias database obtaining specificity and sensitivity up to 98%.

Real-time discrimination of multiple cardiac arrhythmias for wearable systems based on neural networks

Appropriate organization of video databases is essential for pertinent indexing and retrieval of visual information. This paper proposes a new feature called block intensity comparison code (BICC) for video classification and retrieval. Block intensity comparison code represents the average block intensity difference between blocks of a frame. The extracted feature is further processed using principal component analysis (PCA) to reduce the redundancy while exploiting the correlations between the feature elements. The temporal nature of video is modeled by hidden Markov model (HMM) with BICC as the features. It is found that, BICC outperforms other visual features such as edge, motion and histogram which are commonly used for video classification.

A novel block intensity comparison code for video classification and retrieval

The greatest challenge faced during the process of diagnosis of cardiovascular diseases is the accurate analyses of the Electrocardiogram (ECG). Many researches are being done to classify and analyze the ECG signals automatically. In this paper, a novel method for the Auto analysis of the ECG signals using MATLAB is proposed and implemented. In this method, the raw ECG data obtained from the patient goes through a process of Wavelet Packet Decomposition (WPD) followed by Feature extraction. The classification is further done using Artificial Neural Network (ANN). This method, succeeding in differentiating the Abnormal ECG signals from the Normal signals, is proved to be a novel method for Auto analysis of ECG signals.

Auto analysis of ECG signals using artificial neural network

This paper presents investigations about the automatic identification of video genre by audio channel analysis. Genre refers to editorial styles such commercials, movies, sports... We propose and evaluate some methods based on both low and high level descriptors, in cepstral or time domains, but also by analyzing the global structure of the document and the linguistic contents. Then, the proposed features are combined and their complementarity is evaluated. On a database composed of single-stories web-videos, the best audio-only based system performs 9% of Classification Error Rate (CER). Finally, we evaluate the complementarity of the proposed audio features and video features that are classically used for Video Genre Identification (VGI). Results demonstrate the complementarity of the modalities for genre recognition, the final audio-video system reaching 6% CER.

Audio-based video genre identification

Automatic classification of video content is receiving increased impact in the multimedia information processing. This paper inspects the problem of automatic video classification using static and dynamic features. Five different genres such as cartoon, sports, commercials, news and TV serial are studied for assessment. The approach exploits edge information and color histogram as static features and motion information as the dynamic feature with hidden Markov model (HMM) as the classifier. The results are evaluated by constructing individual HMM for each of the features and finally the results obtained are combined to assess the output genre. The method demonstrates the efficiency of the system by applying it on a broad range of video data: 3 hours of video is used for training purpose and a further 1 hour of video as test set. Overall classification accuracy of 95.6% is accomplished.

M.K. Geetha

Papers

HMM Based Automatic Video Classification Using Static and Dynamic Features