Neuropixels 1.0-NHP Long: High-Density Neural Probe

• The probe significantly advances electrophysiological recording by achieving single-neuron resolution across multiple cortical layers and deep brain regions.
• It features 4,416 recording sites along a 45 mm shank with a 103 μm axial pitch, offering ~8-fold denser sampling than traditional arrays.
• Clinical applications demonstrate its safety and efficiency for intraoperative use, enabling high-resolution neural data acquisition in both research and neurosurgical procedures.

Neuropixels 1.0-NHP Long is a silicon-based, high-density, single-shank electrode probe engineered for electrophysiological recording in large-brain mammalian systems. Originally purposed for non-human primates (NHP), its 45 mm length facilitates single-neuron resolution recording across multiple cortical layers and deep brain structures simultaneously. Recent reporting of successful intraoperative deployment in humans has established new standards in high-density, multi-depth neural sampling, spatial resolution, temporal precision, and operational safety in the context of neurosurgical procedures (Brown et al., 14 Jan 2026).

1. Physical and Electrical Specifications

Neuropixels 1.0-NHP Long is fabricated as a monolithic silicon shank, optimized for penetration and stability in large-brain tissue. Geometrically, the probe has a shank length $L$ of 45 mm, a thickness $T$ of 122 μm, and width $W$ of 125 μm. There are 4,416 recording sites along the shank, with 384 channels electrically selectable for simultaneous recording. Each contact has an area $(A_e)$ of $12~\mu m \times 12~\mu m = 144~\mu m^2$ and is arranged with an axial pitch $\Delta x = 103~\mu m$ and a transverse pitch of $20~\mu m$. Typical contact impedance $(Z_e)$ is 150–250 kΩ at 1 kHz, supporting signals between 0.5 Hz and 10,000 Hz, sampled at $f_s = 30,000$ samples/s per channel. The spatial resolution $\Delta x$ of $103~\mu m$ enables a ~8-fold denser sampling than Utah arrays and ~7.7-fold compared to conventional sEEG contacts (800 μm diameter).

The signal-to-noise ratio (SNR) on a per-contact basis is defined as $$\mathrm{SNR} = \frac{\mu_{\mathrm{signal}}}{\sigma_{\mathrm{noise}}}$$, where $\mu_{\mathrm{signal}}$ is the average spike amplitude and $\sigma_{\mathrm{noise}}$ the background noise standard deviation (measured 5–15 dB in human recordings). Empirically, median single-unit SNR in human cortex is 7.5 (IQR 5.2–10.3), a marked improvement over microwire bundles (3–5) (Brown et al., 14 Jan 2026).

Parameter	Value	Note
Shank length $(L)$	45 mm	Optimized for primate and human brains
Total recording sites	4,416	Single-shank distribution
Simultaneous channels	384	User-selectable
Axial pitch $(\Delta x)$	103 μm	Spatial resolution along shank
Sampling rate $(f_s)$	30,000 samples/s	Submillisecond temporal resolution
SNR (median, cortex)	7.5 (IQR 5.2–10.3)	Compared to 3–5 for microwires

2. Surgical Implementation and Custom Instrumentation

A specialized intraoperative workflow is required to ensure mechanical precision, electrical integrity, and sterility. Two to three days prior to surgery, each probe undergoes electrical validation with OpenEphys self-tests. During tray modification, probes are immobilized in sterile, custom-recessed trays adapted for the shank and ribbon cable, minimizing mechanical stress through sterilization cycles.

Insertion guidance and stability are achieved via a 3D-printed Surgical Guide-resin stereotactic holder, featuring anti-rotation channels and 1 mm depth markings. Depending on neurosurgical context, guide tubes (for robot-assisted craniotomies, e.g., Globus ExcelsiusGPS) or custom guide blocks and “straws” (for CRW-frame DBS) direct and protect the probe and associated wiring.

All 3D-printed instrumentation is EO-sterilized as Class 1 Surgical Guide devices. Non-sterilizable electronics (headstage, ribbon cable) remain separated from the sterile field using pouching techniques with inverted glove “thumbs” and probe covers. The complex is clamped into the surgical rig so that the probe enters tissue with collinear and tension-free wiring, minimizing mechanical artifacts (Brown et al., 14 Jan 2026).

3. Intraoperative Insertion Protocol and Noise Mitigation

During neurosurgery, a small dural incision (durotomy) is performed: within a conventional craniotomy or via burr-hole craniostomy. The probe is advanced at ~1 mm/s to depths of 5–40 mm, controlled visually by etched metric markings. Multiple intracranial regions—including postcentral gyrus, middle frontal gyrus, hippocampus, and cingulate cortex—can thus be accessed without dedicated trajectories for each target.

To optimize signal quality, extensive line-noise mitigation steps include disabling or operating all electronics (phones, electrocautery, robotics, infusion pumps) on batteries where possible. Reference and ground needle electrodes are checked every 5 min for secure placement; irrigation with sterile saline prevents desiccation and shorts.

During 30-minute research epochs, intraoperative teams monitor for hemodynamic instability, seizure induction, or probe fracture. No intraoperative probe breakage or neurological events occurred in nine recorded patient cases. Postoperative imaging (where performed) indicated no hemorrhage along probe tracts (Brown et al., 14 Jan 2026).

4. In Vivo Recording Performance and Benchmarking

Neuropixels 1.0-NHP Long demonstrated multi-laminar and deep structure coverage, with recording spans of ~3.8 mm to 30 mm (from 5 to 40 mm depth). High-yield single-unit isolation (10–50% of active channels, via Kilosort4 with manual curation) was achieved. In a representative hippocampal penetration (span 7.7 mm), 25 single units were isolated (yield ≈3.2 units/mm), considerably exceeding macro-electrode sEEG performance (≤1 unit/mm).

The 30 kHz sampling rate, 0.5–10 kHz bandwidth, and low noise permitted submillisecond spike detection and robust spike time cross-correlations (<0.1 ms jitter) across depth. Multi-region synchrony was identified during contralateral DBS, e.g., cingulate-motor cortex activity coherent over tens of milliseconds. This spatial and temporal acuity surpasses conventional Utah arrays and microwire bundles, both in density and simultaneous cross-structural access.

Modality	Axial Contact Pitch	Median Single-Unit SNR	Coverage	Simultaneity
Neuropixels 1.0-NHP Long	103 μm	7.5 (IQR 5.2–10.3)	Surface ⟶ Deep	Multiple regions, single pass
Utah Array	400 μm	~3–5	Surface	Single cortical plane
sEEG	800 μm (diameter)	≤1 unit/mm	Deep	One region per electrode track

5. Safety Outcomes and Clinical Feasibility

Across nine neurosurgical patients (epilepsy, tumor resection, DBS), deployment of the Neuropixels 1.0-NHP Long incurred no observed intraoperative seizures, hemorrhage, probe fractures, or postoperative neurological deficits. Intraoperative monitoring, combined with hardware and protocol measures for electrical and physical integrity, ensured safe, stable recordings. Postoperative imaging, where indicated, revealed no adverse tissue effects along probe insertion tracts. All subjects recovered along standard clinical trajectories, supporting the safety and feasibility of this approach for research and clinical neurophysiology (Brown et al., 14 Jan 2026).

6. Applications and Outlook

Immediate research applications include investigation of memory and language encoding within hippocampal and temporal lobe structures (including awake speech paradigms), the study of neural circuit dynamics during intraoperative DBS (spike timing and evoked potential analysis), and high-density mapping of epileptogenic cortex.

Future directions involve chronic implantation to assess long-term tissue response, execution of closed-loop neuromodulation trials, and the integration with optical or pharmacological perturbation techniques. The publicly available 3D-printed instrument designs (Zenodo 9908900) and detailed, modular sterilization protocols are intended to support multi-center adoption and facilitate rapid advances in high-resolution human systems neuroscience and precision neurosurgery (Brown et al., 14 Jan 2026).