BCI Signal Acquisition

Signal acquisition is the hardware foundation of every brain-computer interface — how neural activity is captured before any decoding can occur. The field is structured around a three-tier invasiveness spectrum: non-implantation, intervention (semi-invasive), and implantation. Each tier makes a distinct trade between signal fidelity and surgical risk.

Non-invasive (non-implantation): EEG is the dominant modality — electrodes on the scalp capture aggregate electrical activity from millions of neurons. fNIRS (functional near-infrared spectroscopy) measures hemodynamic responses as a proxy for neural activity. These methods are safe, portable, and suitable for consumer and rehabilitation applications. The fundamental limitation is that the skull and scalp attenuate and blur neural signals, limiting spatial resolution to ~centimeter scale and temporal resolution is good but signal-to-noise is poor.

Minimally invasive (intervention): This tier includes endovascular approaches like Synchron's Stentrode, in-ear bioelectronics, and other semi-invasive technologies. These achieve better signal quality than surface EEG while avoiding full craniotomy. The Stentrode lodges in a cortical blood vessel and records local field potentials — significantly more informative than scalp EEG. Flexible neural interfaces conforming to brain surface without penetrating tissue (ECoG) also occupy this middle tier.

Fully invasive (implantation): Intracortical electrode arrays like Neuralink's N1 (1,024 electrodes) penetrate brain tissue to record individual neuronal spikes. This yields the highest-bandwidth, highest-fidelity signals: single-unit spiking activity that captures the actual computational primitives of the brain. Signal quality sufficient to decode fine motor intention, speech phonemes, and complex cognitive states. The cost is craniotomy, long-term electrode biocompatibility challenges, and regulatory barriers.

The critical insight from the CUHK Shenzhen comprehensive review (arXiv 2503.16471) is that paradigm design and acquisition technique are not independent — they must be co-designed. SSVEP paradigms (visual stimulation) work best with posterior EEG channels. Motor imagery paradigms require coverage over motor cortex. P300 paradigms need temporal precision that EEG can provide. Mismatching paradigm to acquisition modality wastes the signal information available.

Key Claims

• Three-tier invasiveness spectrum structures the BCI field — Non-invasive (EEG/fNIRS) → intervention (Stentrode, ECoG) → implantation (intracortical arrays). Signal quality scales with invasiveness. Evidence: strong (CUHK Review)
• EEG spatial resolution is fundamentally limited by skull volume conduction — Millimeter-scale neural patterns spread to centimeter-scale on scalp. Deep brain activity is inaccessible. Evidence: strong (CUHK Review)
• Paradigm and acquisition modality must be co-designed — SSVEP requires posterior channels; motor imagery requires sensorimotor cortex coverage; P300 requires temporal precision. Evidence: strong (CUHK Review)
• Emerging intervention technologies balance invasiveness and signal quality — Stentrode, in-ear bioelectronics, flexible ECoG closing the gap between EEG and intracortical. Evidence: moderate (CUHK Review, Clinical Review)
• Flexible neural interfaces reduce signal degradation — Conformal electrode materials minimize motion artifacts and improve skin/tissue contact. Evidence: moderate (Clinical Review)

Acquisition Modality Comparison

BCI Interaction Paradigms

Tied to acquisition modality, paradigms define how users generate control signals:

• Motor Imagery (MI): User imagines movements (left/right hand, feet). Decodes sensorimotor rhythms (mu/beta waves). Requires training; performance varies across individuals. Works with EEG and intracortical.
• P300 Evoked Potential: User attends to rare targets in stimulus sequences. Strong evoked response ~300ms post-stimulus. High accuracy potential, rapid user adaptation. Works well with EEG.
• SSVEP (Steady-State Visual Evoked Potentials): User fixates on flickering stimuli at specific frequencies; occipital cortex generates matching frequency responses. Robust and fast but causes visual fatigue with prolonged use.
• Hybrid Paradigms: Combining MI + P300 or MI + SSVEP to improve robustness and accuracy.
• Emerging Paradigms: Auditory BCIs (attention to auditory streams), olfactory BCIs, tactile BCIs, passive BCIs (state monitoring for fatigue, attention, emotion).

Open Questions

• Can flexible endovascular electrodes approach intracortical signal quality without penetrating brain tissue?
• What is the theoretical information rate ceiling for non-invasive EEG-based BCIs?
• How do multi-day recording stability and signal drift affect long-term implant utility?
• Can emerging paradigms (auditory, olfactory) reliably work for patients with visual impairments?
• What materials science advances will extend intracortical electrode longevity beyond current 5-year estimates?

Related Concepts

• Invasive vs. Non-Invasive BCI — Framing the signal quality/risk tradeoff
• Neural Decoding — How acquired signals are translated to commands
• BCI Clinical Applications — What the signals are used for

Changelog

• 2026-04-14 — Created from CUHK Shenzhen review (arXiv 2503.16471) and ScienceDirect clinical review. Covers full three-tier invasiveness taxonomy, paradigm overview, and acquisition modality comparison table.