This paper proposes a deep learning framework F-GCN that integrates multiple wavelet bases, and extracts MI brain electrical features based on the functional topological relationships between electrodes. The average accuracy of the fused features reaches 92.44%, which is significantly higher than that of a single wavelet basis (coif4: 67.67%, db4: 82.93%, sym4: 73.10%). It also demonstrates good stability and individual convergence in the leave-one-out verification, proving the effectiveness of the method.

Researchers from the Yong Loo Lin School of Medicine, National University of Singapore (NUS Medicine) and National Center for Nanoscience and Technology, the Chinese Academy of Sciences (CAS), have jointly developed an approach to significantly reduce the risk of cancer recurrence and metastasis after surgery, by targeting both bulk cancer cells and the elusive cancer stem cells (CSCs) responsible for relapse.