Brain-computer interfaces provide innovative solutions for mental health monitoring, particularly in depression detection by employing electroencephalography (EEG) signals. This research presents Wave-SpatialNet, a deep learning model tailored to analyze EEG data obtained from only two electrodes, Fp1 and Fp2, to identify patterns associated with depression. The model is highly effective in capturing spatial, temporal, and frequency domain features, providing a detailed insight into brain activity patterns. The architecture of WaveletNet consists of two primary components: one focused on spatial feature extraction from the Fp1 and Fp2 electrodes, and the other dedicated to capturing temporal and frequency features through wavelet transformation. This dual approach allows for detailed analysis of EEG signals, enhancing the model’s capability to differentiate between brain patterns indicative of depression and those observed in healthy individuals. By utilizing only Fp1 and Fp2, our experiments demonstrate that effective depression detection can be achieved with a minimal number of electrodes, significantly reducing complexity and cost while increasing the practicality of the BCI system for real-world applications.