Accurate detection of cardiac arrhythmias from electrocardiogram (ECG) signals plays a vital role in computer-aided diagnosis systems. This study presents an enhanced arrhythmia classification framework based on Discrete Wavelet Transform (DWT) and Multilayer Perceptron (MLP) Neural Networks (NN). ECG recordings are decomposed into multiple frequency bands using DWT, and a set of statistical descriptors is extracted from the resulting wavelet coefficients to characterize signal behavior in both time and frequency domains. These descriptors are integrated with ECG morphological information and heartbeat interval features to construct a comprehensive feature representation. To eliminate redundant information and improve classification efficiency, a feature selection stage is employed before the classification process. The selected features are then used to train an optimized MLP classifier for distinguishing normal heartbeats from different arrhythmia types. Experiments conducted on recordings from the MIT-BIH Arrhythmia Database demonstrate that the proposed combination of wavelet-based statistical analysis, feature optimization, and neural-network classification provides reliable and accurate arrhythmia recognition performance.