Restoring Sight via Photovoltaic Pixels: Inside Science Corp's PRIMA Subretinal BCI for Stargardt Disease

Retinal Prosthetics Enter the Clinical BCI Era: Beyond the Cortical Monopoly

At the TechCrunch Disrupt conference in San Francisco, Science Corporation Chief Executive Officer Max Hodak revealed the first human clinical results from the company’s subretinal neuroprosthetic trial for Stargardt disease. The patient, treated in May at the Sydney Eye Hospital in Australia, achieved letter recognition during their first rehabilitation session and read complete words by session two.

The achievement marks a critical pivot in how the tech industry conceptualizes brain-computer interfaces (BCIs). While companies like Neuralink dominate public attention with penetrating microelectrodes drilled into the motor cortex to control computer mice, Science Corporation has targeted a different entry point into the human nervous system: the retina. Embryologically, the retina is not a peripheral sensory organ; it is an exposed, direct extension of the central nervous system.

Stargardt disease is an inherited genetic macular degeneration that affects roughly 150,000 individuals worldwide. The condition selectively destroys the light-sensing photoreceptor cells in the outer retina, generating an expanding central blind spot (scotoma) that robs patients of the ability to read, identify faces, or perform focused manual tasks. However, the inner retina—specifically the bipolar and ganglion cell layers that process electrical impulses and pipe them down the optic nerve—frequently remains structurally intact and biologically receptive.

Hodak’s thesis treats vision as an electrical signaling protocol. Photoreceptors act as biological solar cells that convert incoming light photons into millivolt-scale electrical potentials. If an engineered microchip can replicate that transduction mechanism at the base of the retina, the optic nerve accepts the artificial stimulation and routes it directly to the brain’s visual processing centers. By validating this principle in an incurable juvenile genetic disease, Science Corporation has established that neural interfaces can restore sensory inputs, rather than merely decoding motor outputs.

Engineering Specifications and Clinical Metrics: Photovoltaic Pixels in Practice

The core hardware in this clinical milestone is the PRIMA system, a modular neuroprosthetic platform designed to operate inside the fluid dynamics of the eye without requiring physical wires or external transcutaneous cables.

The implant itself is an autonomous, passive 2-millimeter by 2-millimeter silicon chip, approximately 30 microns thick. The wafer features an array of 378 photovoltaic pixels arranged in a hexagonal grid. Each pixel contains integrated photodiodes and a central platinum electrode. Unlike older retinal implants, such as the Second Sight Argus II, PRIMA requires no physical power cords connecting the eye to behind-the-ear batteries, eliminating the primary cause of long-term ocular infection and mechanical failure.

Instead, power and visual data arrive via an optical wireless link. The patient wears a specialized pair of glasses containing an outward-facing video camera and an eye-facing near-infrared (880-nanometer) digital micro-projector. A pocket processing unit computes edge detection, contrast enhancement, and digital zoom. It then projects invisible pulsed infrared patterns through the pupil directly onto the subretinal microchip. The silicon pixels absorb this infrared energy, convert it into localized electrical current, and stimulate the adjacent bipolar neurons.

Data from the European PRIMAvera clinical trials—conducted across 38 patients suffering from geographic atrophy (dry age-related macular degeneration)—demonstrated sustained functional efficacy over 36 months. Patients who mastered the system went from total central visual acuity loss to reading standard text sequences on eye charts, with an average improvement of multiple lines of acuity. The successful transfer of this system to Stargardt disease confirms that the photovoltaic array can interface with juvenile retinal neurobiology just as effectively as aged tissue.

Operational and Clinical Realities: How Subretinal Implants Alter Care Delivery

Deploying a physical computing array inside the ocular chamber introduces practical clinical trade-offs that contrast sharply with purely software-driven or drug-based therapies.

Surgical Simplicity and Outpatient Operating Room Economics

Traditional neurosurgery requires an extensive multidisciplinary team, long hospitalizations, and dedicated operating rooms. In contrast, PRIMA is implanted via standard pars plana vitrectomy—a routine, 60-minute outpatient procedure performed thousands of times daily by vitreoretinal surgeons.

Surgeons make a sub-millimeter retinotomy incision, inject a localized balanced salt solution to raise a shallow retinal detachment bleb, slide the 2-millimeter chip into the subretinal space, and flatten the retina back over the chip with silicone oil or intraocular gas tamponade. This fits seamlessly into existing eye hospital workflows without requiring multi-million-dollar surgical robotics.

Patient Rehabilitation Lead Time and Functional Adaptation

Restoring biological signals to the optic nerve does not instantly recreate normal sight. Because the PRIMA chip generates sharp black-and-white visual signals across a central 2-millimeter area, patients describe the initial experience as looking through a narrow straw. The visual input lacks color and peripheral field integration.

Clinical centers must pair surgery with occupational rehabilitation. Patients must learn eccentric fixation and coordinated head-scanning movements:

• First sessions train visual cortex interpretation of high-contrast pixel arrays into discrete geometric shapes.
• Subsequent sessions expand this decoding into character parsing and word construction.
• Patients use optical zoom controls on the processing unit to adjust magnification depending on ambient text size.

Regulatory Precedents and the European Commercial Bridge

Medical hardware commercialization faces distinct regulatory environments across major markets. PRIMA holds commercial approval (CE mark) across the European Union and supply registration in the United Kingdom for treating geographic atrophy. This commercial footprint provides a direct source of real-world patient registries, reimbursement precedents, and post-market safety data.

In the United States, however, commercial clearance through the Food and Drug Administration (FDA) is proceeding through investigational device exemption channels, with broader clearance not anticipated before 2028. This gap is driving clinical tourism, with early trials and private commercial implantation concentrating in Australian, British, and European clinical centers.

Photovoltaic Arrays vs. Direct Cortical Penetration: The Structural Safety Buffer

The broader neural engineering landscape is divided between two philosophies: direct intracortical penetrating arrays and minimally invasive peripheral-central interfaces.

Science Corporation’s subretinal strategy sidesteps the primary bioengineering hazard of neurotechnology: foreign body responses in deep cortical tissue. When microelectrode arrays penetrate the brain’s parenchyma, the brain’s immune system deploys microglial cells to encapsulate the foreign material. This glial scar tissue creates an electrical insulation barrier that degrades signal quality over 12 to 24 months.

PRIMA circumvents this issue through two fundamental design choices:

1. Passive Subretinal Architecture: The implant sits in the subretinal space, an immune-privileged environment with exceptional long-term hardware tolerance. Data from the 36-month PRIMAvera follow-up revealed zero chronic degradation in signal transfer; in fact, visual function metrics at month 36 exceeded those observed at month six, as patient neural plasticity adapted to the artificial signal stream.
2. Surface Cranial Roadmap: For brain conditions outside the eye—such as stroke rehabilitation and brain tumors—Science Corporation is not using penetrating cortical threads. Instead, their experimental neural interface sits on the surface of the cerebral cortex, beneath the skull. By avoiding parenchymal tissue penetration, the interface minimizes vascular damage and microhemorrhaging while still capturing broad cortical field potentials.

Next-generation iterations of the PRIMA platform, scheduled for human trials within the coming year, directly address current resolution limitations. Hodak’s engineering team is manufacturing flexible, contoured silicon substrates that match the natural curvature of the human posterior pole. By reducing electrode spacing and shrinking pixel geometry, these upcoming chips aim to expand the visual field of view from a narrow straw to a broader field, targeting near-native visual acuity over the next five years.

The Commercial Neurotech Horizon: Market Realignment and Strategic Imperatives

Science Corporation’s expansion from age-related macular degeneration to genetic conditions like Stargardt disease demonstrates that retinal neural interfaces can successfully scale across multiple ocular pathologies. This shifts neuroprosthetics from experimental laboratory prototypes to a viable clinical specialty.

How Legacy Vision Therapeutics Face Severe Margin Compression

The validation of subretinal electronics places traditional ophthalmic therapeutic roadmaps under pressure:

• Gene Therapy Limitations: Gene therapy products (such as voretigene neparvovec) command prices exceeding $850,000 per patient and require functional, surviving photoreceptor cells to be viable. In advanced Stargardt disease or end-stage geographic atrophy, those target cells have already died. Gene therapy cannot rescue tissue that is no longer there.
• Biologic Deceleration: Complement inhibitors (such as pegcetacoplan and avacincaptad pegol) merely slow down the expansion rate of geographic atrophy lesions. They do not restore visual acuity, nor can they return the ability to read to a patient who has already lost central vision.
• The BCI Advantage: Hardware neural prostheses are pathology-agnostic at the photoreceptor layer. Whether light receptors died due to a genetic ABCA4 mutation (Stargardt) or senile oxidative stress (macular degeneration), the photovoltaic implant replaces their bioelectric output entirely.

Three Non-Negotiable Benchmarks for BCI Commercial Dominance

For medical device enterprises, health systems, and neurotech investors tracking the evolution of neural interfaces, three strategic criteria dictate long-term operational viability:

• Hardware Independence from Active Internal Power: Neuro-implants that rely on internal chemical batteries face strict shelf lives and thermal dissipation ceilings that limit computing power. Future interface leaders will rely on optical or RF passive power harvesting, offloading battery weight, thermal risk, and processing tasks to external wearable devices.
• Surgical Alignment with Existing Specialty Ecosystems: Interfaces that require custom neurosurgical operating rooms or proprietary installation robots will remain niche, high-cost therapies. Hardware designed to leverage standard surgical tools (such as 25-gauge vitrectomy cutters and common ophthalmic microscopes) can scale across thousands of hospital networks worldwide.
• Rehabilitation Workflows Paired with Machine Vision: The hardware implant is only one component of the sensory loop. The ultimate visual performance of systems like PRIMA depends on real-time computer vision running on the external processor—using edge AI to outline high-contrast obstacles, zoom into text, and convert messy environments into structured electrical pulses that the patient’s visual cortex can parse.