<p>With the recent increase in monkeypox outbreaks, more sophisticated diagnostic methods are needed. Traditional approaches suffer from a lack of precision and generalizability, whereas state-of-the-art machine-learning techniques can be exploited with better detection and classification. Our proposed method uses novel optimization strategies to improve feature extraction and significantly outperform existing models, indicating it has real-world applicability in clinical settings. This manuscript proposed a new approach toward the detection of Monkeypox Virus Disease: feature extraction along with the application of Conditional Improved Wasserstein Generative Adversarial Networks, CIWGAN, to enhance the diagnosis accuracy. One major innovative aspect is the use of a hybrid optimization technique employed within the context of Slime Mould Optimization (SMO) in combination with Garra Rufa Optimization (GRO) within the Adaptive and Concise Empirical Wavelet Transform (ACEWT) weight parameters. In other words, while ACEWT efficiently extracts the statistical and geometric features from the skin lesion images, the proposed adaptive optimization significantly improves its performance. First, the proposed work used a skin lesion dataset, where noise in the input images was removed by Fast Guided Filter (FGF), followed by feature extraction of the noise-reduced images using ACEWT later followed by optimization with the above proposed hybrid technique. Feed the optimized features into CIWGAN to classify the samples as Monkeypox or normal. The proposed method thus showed an impressive increase in accuracy: 13.14%, 11.02%, and 9.11% better than any other methods. The explainable AI (EAI) analysis, which also helps understand the decision-making process after classification, focuses on the most relevant features affecting the model's predictions. The execution is on the MATLAB platform, hence showing that the proposed approach would be very effective in pinpointing the occurrence of Monkeypox with much accuracy.</p>