چكيده به لاتين
Cardiovascular disease is one of the most important causes of mortality in the world and the first one in Iran. One of the cardiovascular medical devices is Automatic defibrillator that automatically detects the heart attack and recovers the proper performance of the heart by applying voltage to the heart. In this study, proper pre-processing including noise removal, baseline removal, amplitude normalization and sampling frequency normalization are applied to the ECG signal, then PQRST points are extracted and features of Atrial fibrillation arrhythmia are generated. By calculating the Likelihood function for each feature, we evaluated the features and selected the best ones. Then, using the FLD method, we reduced the dimension of the feature space, once from eight selected features and once from four selected features to one dimension. To classify the data using these features, three methods are used, Artificial neural networks, Bayes and Random forest. All of the above steps, from pre-processing to data classification, after programming in MATLAB, are converted to C++ and implemented on the STM32F407VGT6 processor that uses the ARM architecture. Proper PCB is designed and implemented and the results are visible through both on-board LEDs and the TFT LCD. In addition, data is stored on the SD-Card and the device is able to receive the input signal through USB, SPI, UART and USART. The implemented circuit is capable of correctly classifying 80% of the validation data using only one feature.
