Epilepsy is a neurological condition characterized by unpredictable disturbances in brain activity, leading to disruptive epileptic seizures. Despite its high prevalence with over 50 million individuals affected globally, epilepsy detection remains challenging, mainly due to limitations in conventional analysis of EEG signals. This study aims to develop a wavelet entropy (WE) method from one-dimensional signals to 2D on spectrogram images of EEG seizure signals, focusing on deeper analysis and complex pattern detection. The proposed problem-solving approach is to manage the entropy values ??generated from the Discrete Wavelet Transform (DWT) process to obtain wavelet entropy values. This aims to obtain important features that are not always clearly identified through direct analysis of the spectrogram. The advantage of this study is the use of the WE method developed from one-dimensional signals to 2D images, utilizing information in the spectrogram for more complex pattern detection. The research roadmap includes the initial stages of reading the EEG seizure dataset, converting signals into images through Short Time Fourier Transform (STFT), and using wavelet entropy for feature extraction. The classification stage using deep learning is expected to recognize complex patterns in the extraction of spectrogram images. The accuracy results obtained are quite high, which is 100% at level 7 using several types of mother wavelets. Thus, this research can be an alternative for advanced 1-dimensional signal processing.