Visual Field Analysis With Artificial Intelligence

RESEARCH PROTOCOL

Study Title:

Optimizing Visual Field Analysis using Artificial Intelligence: Establishing Normal Threshold Values for Healthy Participants Using a Visual Field Perimeter

Principal Investigator:

Edward Chay, MD

Site Location:

Sight Intelligence Engineering Corporation 8101 Hinson Farm Road, Suite #208 Alexandria, VA 22306

IRB Approval:

Approved by the Institutional Review Board (IRB) at INOVA

1. Background and Rationale Visual field testing is a key diagnostic tool in ophthalmology for detecting and monitoring conditions such as glaucoma and optic neuropathies. This study aims to establish normative threshold values for a new visual field perimeter in a population of healthy individuals. These baseline values will contribute to the calibration and optimization of a visual field perimeter system that uses an advanced algorithm, potentially incorporating Artificial Intelligence (AI), to enhance the accuracy and efficiency of visual field testing.
2. Objectives

• To collect normal visual field threshold values in a retrospective chart review, then test the developed analyzer software in a prospective study using healthy individuals.
• To assess usability and comfort of the visual field perimeter device.
• To support optimization of the perimeter algorithm, including AI-driven test point selection.

1. Study Design Part one involves a brief retrospective chart review to gather data to build an AI model differentiating between healthy and glaucomatous visual fields. Part two is a prospective, single-site observational study involving 100 healthy participants. Each participant will complete a single visual field test using a virtual reality-based perimeter.
2. Participant Selection Criteria Inclusion Criteria:

• Healthy adults aged 18 years or older.
• Best corrected visual acuity of 20/40 or better in each eye.
• Normal optic nerve appearance upon clinical examination. Exclusion Criteria:
• Best corrected visual acuity worse than 20/40.
• Cup-to-disc ratio greater than 0.5.
• Previously documented glaucoma or optic neuropathy.
• Any significant ocular or neurological condition that may impact visual field.
• Inability to comply with study procedures.

1. Recruitment and Consent Participants will be ethically recruited from Dr. Chay's established patient base at Schefkind Eye Care, following the recruitment procedures described in a separate IRB-approved document. Dr. Chay will not directly approach his patients to minimize any potential for coercion. Upon arrival at the study site, participants will receive a brief introduction to the study. Written informed consent will be obtained by a member of the research team not involved in clinical care. A copy of the signed consent form will be provided to the participant.
2. Study Procedures

• A de-identified database of patients previously seen at Schefkind Eye Care will be split into healthy and glaucomatous groups by single-blinded physician review. The visual field test outputs will be used to train an AI model to perform the same splitting function.
• In the prospective portion of the study, informed consent is obtained, participants will be fitted with the virtual reality headset that runs the AI model.
• The headset may be adjusted by the test administrator for comfort and visual alignment.
• Instructions for the visual field test will be displayed within the device in the participant's preferred language.
• The participant will complete the visual field test while the administrator remains in an adjacent room and available at all times to respond to any questions or concerns.
• After the test, participants will receive a verbal debrief and be asked to complete a brief exit survey evaluating their experience.

1. Sample Size A total of 100 participants will be enrolled. This number is sufficient to establish a normative dataset for healthy eyes, allowing for future comparison to pathological cases.
2. Data Analysis Methods Data collected from the visual field tests will include threshold sensitivity values at various visual field points. The dataset will be analyzed using descriptive statistics (mean, standard deviation, range) to determine normal threshold ranges across test locations. Additional analysis may include:

• Stratification by age group and gender to identify demographic influences.
• Outlier detection to ensure data quality.
• Internal consistency analysis of test-retest reliability if applicable. All analyses will be performed using statistical software such as R or Python. De-identified data may be used in the future for algorithm training or model development.

1. Timeline Phase Activity Duration Dates Phase 1 IRB approval and study preparation 1 month Completed Phase 2 Initial retrospective data analysis + Participant recruitment and enrollment 1 month [January-April 2026] Phase 3 Data collection and visual field testing 1 months (overlapping) [April-May 2026] Phase 4 Data analysis and reporting 1-2 months [June 2026] Phase 5 Final reporting and manuscript preparation 1 month [July 2026] The total anticipated duration of the study is 4-5 months.
2. Data Management and Confidentiality All study data will be stored in a secure, password-protected electronic database. Each participant will be assigned a unique study ID to de-identify personal information. Only authorized study personnel will have access to the data. Data will be handled in accordance with HIPAA and institutional privacy regulations.
3. Risks and Benefits Risks:

• Minimal: potential for temporary eye fatigue or mild discomfort from the visual field test or wearing the headset. Benefits:
• No direct benefit to participants.
• The study contributes to the development of improved diagnostic tools and may benefit future patients through more accurate visual field testing.

1. Compensation Participants will not receive financial compensation for participation.
2. Voluntary Participation Participation in the study is entirely voluntary. Participants may withdraw at any time without any penalty or effect on their clinical care.
3. Funding Disclosure This study is self-funded by the Principal Investigator, Dr. Edward Chay. No external funding sources or commercial sponsors are involved.