Precision cortical-array feasibility, 2025

Hettick and colleagues’ study combines three kinds of evidence: minimally invasive delivery in pigs and cadaveric human heads, animal recording and stimulation, and a five-patient intraoperative recording pilot. These are not one chronic human BCI trial.

Delivery and animal work

The authors performed 22 cranial micro-slit insertions in eight Göttingen minipigs. A further 61 arrays were implanted in 34 minipigs for electrophysiology and biocompatibility studies. They also tested insertion in 23 fresh cadaveric human heads. A stylet carried the flexible array into the subdural space and was then removed.

The paper reports reversible explantation after implants lasting up to 42 days in its animal study. Histology found preserved cortical-surface architecture and no systematic differences between sampled regions under thin-film arrays and standard four-contact subdural controls. Long-term human implantation is explicitly still needed to substantiate comparable reversibility and safety.

The paper also shows somatosensory and visual decoding, walking-related activity and cortical stimulation in animals. These do not establish restored vision, walking or communication in patients.

Five-patient human pilot

Two neurosurgical patients under general anesthesia supplied spontaneous recordings and somatosensory-evoked-potential mapping. Three awake patients supplied recordings during language mapping, with arrays over the left superior temporal gyrus, angular gyrus or frontal operculum.

The study detected speech onset versus non-speech intervals from cortical activity. The text gives 79.8% accuracy (95% confidence interval 77.6 to 82.0%), and also describes a 79% result using four minutes of training data with 54 utterances. Figure 5 specifies offline decoding assessed with 100 randomly shuffled train-test samples.

This is binary speech-event detection. It is not word transcription, silent attempted-speech decoding in a person with paralysis, or an at-home communication result.

What remains untested here

The authors state that the thin-film interface was not evaluated as a fully integrated chronic wireless package. The animal behaviour and short intraoperative sessions also limited optimization for specific clinical indications. Five intraoperative patients do not establish long-term implant survival or clinical efficacy.

The paper was published online on 2 October 2025; the publisher PDF is paginated in the June 2026 volume. The online publication date is used in this entry.

Sources

• Hettick M, Ho E, et al. Primary paper. Delivery cohort, human pilot, Discussion and fabrication methods.
• Publisher PDF. Primary affiliation: Precision Neuroscience Corporation, with West Virginia University and Mount Sinai collaborators.
• Figure 1. System and array configurations. Its conceptual panels are not physiological data.