Brain-computer interfaces (BCIs) enable communication by directly decoding neural activity into comprehensible output forms. These interfaces can be broadly categorized into brain-to-text, brain-to-speech, and brain-to-image modalities. Brain-to-text BCIs translate neural activity into written text, facilitating communication without speech or physical interfaces. Brain-to-speech BCIs decode neural signals to synthesize spoken language, preserving personalized features such as tone and prosody, thereby enabling expressive communication. Brain-to-image BCIs reconstruct visual stimuli or imagined images from neural activity, providing a non-verbal channel for expressing complex concepts visually. This review synthesizes recent advancements, identifies ongoing challenges, and highlights the significance of these developments, emphasizing the potential to significantly improve communication and quality of life for individuals with severe motor impairments. Future work should explore minimally invasive implantation techniques, potentially combining benefits of invasive and non-invasive methods, to balance communication accuracy with user safety and accessibility.
Tong Gao (Wed,) studied this question.