The electroencephalogram (EEG) serves as a crucial neurophysiological measurement capturing brain electrical activity through scalp electrodes. As m-healthcare and e-healthcare become indispensable in biomedical science, the challenge of removing artifacts from EEG signals grows. This paper proposes a deep learning enhanced deep hybrid multi-resolution discrete wavelet transform-based delayed error normalized least mean square (MR-DWT-DENLMS) method for eliminating sign artifacts from EEG signals. Additionally, this study will explore the integration of the proposed method with Generalized Frequency Division Multiplexing with Index Modulation (GFDM-IM) for channel estimation. The Singular Value Decomposition-Least Mean Squared Error (SVD-LMMSE) module will be utilized for accurate channel estimation. The performance of the proposed method will be thoroughly evaluated using simulations on the Xilinx platform with Verilog coding. Key performance metrics such as BER and various filter analyses for different parameters.