South Carolina Patent of the Month – January 2026

Comparative Research Report: Systems and Methods for Eyelid Localization (US Patent 12,518,388)

Patent Identification and Overview

The landscape of biometric analysis and neurological diagnostics has been fundamentally altered by the issuance of United States Patent No. 12,518,388, formally titled “Systems and methods for eyelid localization.” This seminal intellectual property, which was applied for on May 27, 2022, and officially awarded on January 6, 2026, represents a pivotal advancement in the integration of machine learning with physiological monitoring. Assigned to BlinkTBI, Inc., a technology firm based in Charleston, South Carolina, the patent details a sophisticated computational architecture designed to track the kinematics of the eyelid with sub-millimeter precision using neural networks. In recognition of its technological novelty and commercial viability, this patent was recently awarded the distinguished title of South Carolina Patent of the Month. This accolade was not the result of a subjective review but was determined through a rigorous, data-driven selection process utilizing advanced Artificial Intelligence (AI) technology, which screened over 1,000 potential patents filed within the jurisdiction to identify the single most promising innovation.

The selection of Patent 12,518,388 as the South Carolina Patent of the Month was predicated principally on its profound real-world impact and its potential to disrupt multiple sectors, from healthcare to law enforcement. While many modern patents describe incremental improvements to existing software or abstract data processing methods, this invention bridges the divide between theoretical computer vision and tangible clinical utility. The AI-driven selection algorithms prioritized this patent because it addresses a critical, unmet need in society: the lack of an objective, portable, and rapid tool for assessing neurological function. By transforming the blink reflex—a biological marker previously measurable only in static, high-cost laboratory settings—into a field-deployable data point, the technology enables immediate diagnostics for Traumatic Brain Injury (TBI), neurodegenerative disease progression, and impairment due to fatigue or intoxication. The patent stands as a technological cornerstone that shifts the paradigm of neurological assessment from subjective physician observation to precise, quantitative reflexometry.

Technical Architecture and Innovation

To comprehend the superiority of the invention described in Patent 12,518,388, it is necessary to deconstruct the technical claims and the underlying engineering challenges it resolves. The patent does not merely describe a camera pointing at an eye; it describes a complex system of “self-healing” visual codes and neural processing designed to function in the chaotic variables of the real world.

The Core Challenge: Eyelid Localization in Uncontrolled Environments

Historically, eye-tracking technology has struggled with the “localization problem.” Traditional algorithms, often based on edge detection or Hough transforms (geometric feature extraction), rely on high-contrast boundaries between the pupil, the iris, and the sclera (the white of the eye). These algorithms function adequately in controlled laboratory environments where lighting is uniform and the subject’s head is immobilized. However, in real-world scenarios—such as a football sideline, a roadside sobriety stop, or a triage tent—these variables are uncontrolled.

• Occlusion: Eyelashes, particularly if they are long or have makeup applied, often obscure the eyelid margin, confusing standard edge-detection algorithms.
• Lighting: Variable ambient light can create shadows in the eye socket, rendering contrast-based detection useless.
• Anatomy: The variability in human facial structure (e.g., deep-set eyes, epicanthic folds) means that a “hard-coded” geometric rule set often fails across diverse populations.

The Solution: Neural Network Integration

Patent 12,518,388 solves these problems through the deployment of a trained neural network comprising at least 10,000 parameters. This represents a shift from “deterministic” programming (telling the computer what an eyelid looks like) to “probabilistic” learning (showing the computer thousands of examples of eyelids and letting it derive the rule set).

• Parameter Depth: The specification of “at least 10,000 parameters” indicates a deep learning architecture capable of capturing subtle, non-linear relationships in the image data. This depth allows the system to distinguish between the actual eyelid margin and a “false positive” like a mascara clump or a shadow.
• Temporal Consistency: The patent describes a method where the inputting and obtaining of coordinates are repeated over a “non-zero duration.” This is critical for kinematic analysis. A single static image of an eye is medically useless for reflex analysis; the diagnostic value lies in the movement—the velocity, acceleration, and latency of the blink. The neural network must, therefore, maintain tracking integrity across a high-speed video frame sequence, ensuring that the “eyelid” detected in Frame 1 is the same anatomical feature detected in Frame 50.

Physiological Significance: The Blink Reflex

The technical innovation is inextricably linked to the physiological mechanism it measures. The technology focuses on the blink reflex, specifically the corneal reflex or the startle response triggered by a puff of air (as utilized in the EyeStat/Blink Reflexometer device).

• Neural Pathway: This reflex is mediated by the trigeminal nerve (Cranial Nerve V), which senses the stimulus, and the facial nerve (Cranial Nerve VII), which drives the orbicularis oculi muscle to close the eye. This pathway traverses the brainstem.
• Diagnostic Value: Because the reflex is processed in the brainstem (a primitive, evolutionarily conserved structure), it is immune to “faking.” A subject can voluntarily slow their eye tracking (affecting cortical pathways), but they cannot voluntarily alter the latency of a reflex arc measured in milliseconds. This makes the patent’s focus on reflexometry technically superior for objective diagnosis compared to technologies that track voluntary gaze.

Comparative Analysis and Benchmarking

The sector of neurological assessment and oculometrics is competitive, occupied by legacy medical device manufacturers and emerging digital health startups. However, Patent 12,518,388 enables a device performance profile that is superior to its competitors. The following analysis benchmarks the patented technology against the three primary categories of competition: Laboratory Motion Analysis Systems (The Gold Standard), Vision Therapy Eye-Trackers, and Standard Clinical Evaluations.

Competitor 1: Laboratory Motion Analysis Systems (e.g., BTS SMART System)

The primary benchmark for scientific accuracy in this field is the SMART Motion Analysis System (BTS, Milan, Italy). This system uses multiple infrared cameras mounted in a dedicated room to track reflective markers placed on the subject’s face.

• Comparison: A validation study compared the portable EyeStat device (utilizing Patent 12,518,388) directly against the SMART system. The results were definitive:
• Correlation: The data showed a linear relationship with correlation coefficients () ranging from 0.57 to 0.85 across various blink parameters.
• EMG Validation: Simultaneous Electromyography (EMG) recordings of the orbicularis oculi muscles showed no statistical difference between the data collected by the two systems.
• Why Patent 12,518,388 is Superior: While the SMART system is accurate, it is immobile, costs hundreds of thousands of dollars, and requires 30+ minutes of setup (placing markers, calibrating cameras). The patented invention achieves equivalent physiological accuracy () in a portable, headset-based form factor that requires no markers and completes a test in under 60 seconds. It democratizes “lab-grade” data.

Competitor 2: Vision Therapy Trackers (e.g., RightEye)

Companies like RightEye and others (e.g., Cerebrotech, NeuroOptics) focus on oculomotor function—tracking the eye as it follows a moving dot on a screen.

• Comparison: These systems measure saccades (rapid jumps) and smooth pursuit (tracking). These are voluntary movements controlled by the frontal eye fields in the cerebral cortex.
• Why Patent 12,518,388 is Superior:
• The “Sandbagging” Problem: In sports and insurance claims, subjects may intentionally underperform on a baseline test (sandbagging) to make post-injury comparison easier, or try to “focus hard” to mask a deficit. Because RightEye tracks voluntary movement, it is susceptible to this cognitive manipulation.
• The Reflex Advantage: The BlinkTBI system measures an involuntary reflex. It is physiologically impossible for a human to voluntarily alter the latency of their brainstem reflex arc by 5 milliseconds. Therefore, the data derived from Patent 12,518,388 is inherently more objective and fraud-resistant than voluntary gaze tracking.

Competitor 3: The Standard Clinical Exam (Penlight/H-Test)

The current standard of care in field settings (sidelines, roadside) is the subjective neurological exam, where a doctor or officer asks the subject to “follow my finger.”

• Comparison: This method relies entirely on the observer’s subjective rating (“The patient looks sluggish” or “The eyes are jerky”).
• Why Patent 12,518,388 is Superior:
• Quantification: The patent enables the measurement of Differential Latency (the time difference between the left and right eye reaction) down to the microsecond. A human observer cannot see a 10-millisecond delay; the human eye operates at roughly 30-60 Hz. The patented system captures at high frame rates, revealing deficits invisible to the naked eye.
• Data vs. Opinion: Dr. Nancey Trevanian Tsai, a key figure in the technology’s development, summarizes the superiority: “It’s analogous to the difference between saying someone is feverish and measuring the patient’s temperature with a thermometer.” Patent 12,518,388 invents the thermometer for the brain.

Real-World Impact and Future Potential

The selection of this technology as the South Carolina Patent of the Month was driven largely by its utility. Unlike theoretical patents, this invention is currently deployed in high-stakes environments and holds the potential to revolutionize public safety.

Traumatic Brain Injury (TBI) and Concussion Management

The “silent epidemic” of concussions affects millions of athletes, soldiers, and accident victims annually. The core problem is diagnosis: structural imaging (CT/MRI) often appears normal because a concussion is a metabolic injury, not a bleeding injury.

• Current Impact: The device is used by sports organizations and the military to establish baselines. When an athlete suffers a hit, they are re-tested. If their blink latency has drifted from their baseline (e.g., from 35ms to 45ms), it provides objective evidence of neural dysfunction, justifying removal from play. This prevents Second Impact Syndrome, a potentially fatal condition where a second concussion occurs before the first has healed.
• Mechanism: The patent’s kinematic analysis detects “number of oscillations” (shuddering of the eyelid) and “latency” (speed of signal transmission), which are direct correlates of the brainstem’s health.

The “Holy Grail” of Law Enforcement: The Marijuana Breathalyzer

Perhaps the most significant future economic potential for Patent 12,518,388 lies in sobriety testing.

• The Problem: With the widespread legalization of cannabis, law enforcement lacks a non-invasive field test. Breathalyzers detect alcohol but cannot detect THC. Blood tests require a warrant and are invasive. Furthermore, THC remains in the body for weeks; a positive blood test does not prove current impairment.
• The Solution: Research utilizing the patented technology has shown that different substances affect the blink reflex in distinct, identifiable patterns.
• Alcohol: generally slows latency.
• Marijuana: creates a specific “jitter” or desynchronization between the eyes.
• Fatigue: affects the velocity of the eyelid opening phase.
• Future Potential: By applying specific AI classifiers (as described in the patent’s neural network claims) to these kinematic signatures, the device can serve as a “Universal Impairment Monitor.” It essentially asks the question: “Is this person’s brain reacting fast enough to safely operate a vehicle?” regardless of whether the cause is Gin, Gummies, or Exhaustion. This application addresses a massive gap in public safety infrastructure.

Neurodegenerative Disease Monitoring

Beyond acute injury and intoxication, the technology has deep applications in chronic disease.

• Parkinson’s Disease: Characterized by dopamine depletion in the basal ganglia, Parkinson’s results in “masked facies” and a reduced blink rate. The system can quantify the rigidity of the eyelid muscles, potentially detecting the disease years before gross tremors appear. It can also track the efficacy of L-Dopa medication by measuring “on” and “off” phases in blink fluidity.
• Huntington’s Disease: The system can detect the early onset of chorea (involuntary movements) in the facial muscles.

R&D Tax Credit Analysis: The 4-Part Test

For deep-tech companies like BlinkTBI, the Research and Development (R&D) Tax Credit is a critical source of non-dilutive capital. It allows companies to offset a portion of the significant costs associated with developing patentable technology. To qualify for the credit under IRC Section 41, the development of Patent 12,518,388 must satisfy the rigorous Four-Part Test.

Swanson Reed, a specialist R&D tax advisory firm, utilizes a distinct methodology to substantiate these claims, ensuring that innovation is rewarded while maintaining strict compliance with IRS regulations.

Part 1: Permitted Purpose

The Requirement: The activity must relate to a new or improved business component (product, process, computer software, technique, formula, or invention) regarding its function, performance, reliability, or quality.

• Project Application: The development of the “Systems and methods for eyelid localization” clearly meets this standard. BlinkTBI sought to develop a new medical device (The EyeStat/Reflexometer) and improved internal software algorithms that could function in uncontrolled lighting conditions—a significant improvement in performance and reliability over existing edge-detection methods.
• Swanson Reed’s Role: Swanson Reed identifies the specific “business components” (the hardware device vs. the software algorithm) to ensure costs are allocated correctly. They ensure that the claim focuses on the technological improvements (accuracy, speed) rather than aesthetic or marketing improvements.

Part 2: Technological in Nature

The Requirement: The research must rely on principles of the hard sciences—physical or biological sciences, engineering, or computer science.

• Project Application: The patent is undeniably technological. It relies on:
• Computer Science: The architecture of Convolutional Neural Networks (CNNs), backpropagation, and parameter optimization (10,000+ parameters).
• Physiology/Biology: Understanding the trigeminal-facial nerve reflex arc.
• Optical Engineering: High-speed infrared image capture and signal processing.
• Swanson Reed’s Role: Swanson Reed utilizes staff with engineering and scientific backgrounds (not just accountants) to review the technical documentation. This “peer-to-peer” review ensures that the claim narrative accurately reflects the hard science, satisfying the “Technological in Nature” test during an audit.

Part 3: Elimination of Uncertainty

The Requirement: The taxpayer must face uncertainty regarding the capability to develop the component, the method of development, or the appropriate design of the component.

• Project Application: At the outset of the project (prior to May 2022), success was not guaranteed.
• Uncertainty of Capability: Could a mobile processor handle the inference load of a 10,000-parameter neural network in real-time (30ms latency)?
• Uncertainty of Design: What is the optimal frame rate? How many IR LEDs are needed to illuminate the eye without causing a corneal reflex from the light itself?
• Uncertainty of Method: How do we train the AI to ignore heavy makeup or dark eyelashes? The existence of the patent itself serves as proof that these answers were not “known” but had to be discovered.
• Swanson Reed’s Role: The firm documents the “technical challenges” faced by the engineering team. They collate emails, bug reports, and meeting minutes that show the team asking “Can we do this?” and “How do we fix this lag?”, which serves as contemporaneous evidence of uncertainty.

Part 4: Process of Experimentation

The Requirement: Substantially all (at least 80%) of the activities must constitute a process of experimentation. This involves identifying uncertainties, identifying alternatives, and evaluating those alternatives (e.g., modeling, simulation, trial and error).

• Project Application: The development history of Patent 12,518,388 is a textbook example of experimentation.
• Hypothesis Testing: The team likely hypothesized that a CNN would outperform Hough transforms.
• Systematic Evaluation: The benchmarking study against the BTS SMART system is a formal evaluation of the alternative. They ran trials on 16 subjects, collected data, analyzed correlations (), and refined the model based on the results.
• Iterative Refinement: The patent claims mention repeating the input “over a non-zero duration,” implying an iterative testing process to tune the temporal tracking parameters.
• Swanson Reed’s Role: This is the most scrutinized area of the R&D credit (especially post-Little Sandy Coal). Swanson Reed ensures that the “Process of Experimentation” is documented not just as “work done” but as a scientific method. They help the client structure their documentation to show the alternatives rejected (e.g., “We tried Algorithm A, it failed in sunlight; we switched to Neural Network B”).

How Swanson Reed Helps Claim the Credit

Swanson Reed provides a specialized service model to secure these claims:

1. Six-Eye Review: Every claim is reviewed by a scientist/engineer, a tax attorney, and a risk manager. This ensures the claim is technically sound (identifying the correct science) and legally compliant.
2. ISO 31000 Risk Management: They employ international risk standards to assess the probability of audit and the strength of the evidence.
3. creditARMOR: An AI-driven audit defense platform that prepares the client for IRS scrutiny before the claim is even filed.
4. Payroll Tax Offset: For a startup like BlinkTBI (if they meet the Qualified Small Business criteria), Swanson Reed can help apply up to $500,000 per year of the credit against payroll taxes, providing immediate cash flow even if the company is not yet profitable.

Final Thoughts

US Patent 12,518,388 is a landmark innovation in the field of biometric analysis. Its recognition as the South Carolina Patent of the Month underscores the successful convergence of Artificial Intelligence and practical medical diagnostics. By leveraging deep learning to achieve laboratory-grade precision in a portable form factor, the invention outperforms existing competitors and opens vast new markets in TBI management and impairment testing. Furthermore, the rigorous development process required to bring this technology to life—marked by technical uncertainty and systematic experimentation—makes it a prime candidate for the R&D Tax Credit. Through the specialized guidance of Swanson Reed, innovators can leverage these tax incentives to fuel the continued advancement of technologies that make our roads safer and our brains healthier.