What Does “Minimally Invasive” Actually Mean?

No regulator defines 'minimally invasive' for brain interfaces. Companies use it to mean that their device avoids something a competitor's does not: opening the skull widely, opening the dura, or penetrating the brain.

A more useful way to compare approaches is by three questions. What is opened? What does the device touch? And what remains in the body?

An endovascular array is delivered by catheter through the jugular vein and expands inside the superior sagittal sinus, beside motor cortex. Nothing is opened in the skull. The procedure draws on techniques that neurointerventional teams already use.

The published record is the strongest of any commercial implant on safety: a first-in-human study followed participants for twelve months. The cost is resolution. The device records through the vessel wall with sixteen electrodes and must go where the veins go.

Thin-film arrays can be slid onto the cortical surface through a slit less than a millimetre wide. The dura is opened, the brain is not penetrated, and the film can be removed.

The channel count is high — 1,024 electrodes per array — and several arrays can be tiled. Human evidence so far is from short intraoperative use and a clearance for up to thirty days. Chronic performance is the open question.

Epidural devices sit in or under the skull without opening the membrane around the brain. Wireless epidural recorders have driven an exoskeleton and formed the cortical half of a brain–spine interface. A newer class of miniature stimulators is seated in the skull for psychiatric indications.

Signals are coarser than on the cortical surface, but the infection barrier stays intact.

Every step away from the neuron costs resolution. Every step toward it costs surgical complexity. The interesting question is not which approach is least invasive but which delivers enough function for a given need at an acceptable risk.

For a person who needs a reliable click, sixteen electrodes may be sufficient. For open-vocabulary speech, the evidence so far comes from denser interfaces.