Abstarct: The recording of the EEG signals could last for several hours, thus the storage of these signals usually occupies a high volume and encountered telemedicine with difficulties. By compression, the volume of the occupied space can be reduced. It is worth noting that the aim of desired compression is removing redundant information from the signal without losing important signal information. In this thesis, a novel method for compression and reconstruction of EEG signals baxsed on wavelet transform and adaptive filter with emphasis on preserving depression recognition performance is proposed. In present study, We first convert the one – dimensional EEG signals into two-dimensional signals with two techniques of array arrangement in order to Zigzag and Spiral and then compressed them with six techniques of wavelet encoder. After compression EEG signals for evaluation of performance of compression by feedforward artificial neural network, we recognize the depression disorder. In the compression section, the evaluation measures we used includes compression ratio (CR), percentage root mean square difference (PRD) and peak signal to noise ratio (PSNR). Moreover, in diagnosis of depression disorder section, we used accuracy, sensitivity, selectivity and specificity evaluation measures. As one of obtained results, in compression section, STW and zigzag methods had the biggest compression performance and in diagnosis of depression disorder SPIHT and zigzag methods had the biggest diagnosis performance.