This study addresses the restricted accuracy and high operational barriers of traditional brain-computer interfaces (BCIs) in high-dimensional control. We propose a novel hybrid BCI fusing Steady-State Motion Visual Evoked Potentials (SSMVEP) and Electromyography (EMG). By aligning a 15-target dynamic graphic zoom paradigm with four dental occlusion EMG patterns, our parallel architecture expands the command space from 15 to 60 targets. Deep learning was employed for multimodal signal decoding, maintaining classification accuracy comparable to unimodal operations while offering an accessible pathway for severe motor impairments. Benefiting from dual-channel processing, the multi-mode Information Transfer Rate (ITR) reached 62.33 bits/min, significantly outperforming the single-mode maximum of 42.49 bits/min. This interference-free mechanism breaks traditional bandwidth ceilings, reducing cognitive load and validating an efficient 60-target control system for complex real-world deployment.