Evaluating a novel AI-based fundus camera device for screening glaucoma in the setting of a comprehensive ophthalmology clinic

The study will be carried out in two stages. Individuals attending the comprehensive ophthalmology clinic who meet inclusion criteria will be asked to participate in the study. The Patient Information Sheet will be provided to them, and after review and discussion, written informed consent will be obtained from those willing to participate in the study. Basic demographic details including age and gender will be obtained from the participants after they undergo refraction and intraocular pressure evaluation by non-contact tonometry.

 Stage 1: Patients will undergo testing with the AI-based fundus camera device before their comprehensive ophthalmology visit. The research investigator will initially attempt to capture the optic nerve head image without dilating the patient, completing a maximum of two attempts to achieve an image of sufficient quality for AI evaluation. It is estimated that up to 40% of patients will require dilation for the device to obtain an adequate, gradable image that allows the Medios AI to assess for glaucoma. If a patient requires dilation, the patient will first be seen by the ophthalmologist who will determine if dilation is safe. After dilation, an additional two attempts will be made to obtain an image of sufficient quality to be graded for glaucoma. If the device is unable to obtain a gradable image following dilation, the patient will be noted to be ungradable by AI and continue with their routine evaluation. The total time taken to obtain the images will be noted down in the proforma.

 Testing with the AI-based fundus camera device offers four possible outcomes:

·        Referable Glaucoma (RG) – The patient requires referral to a glaucoma clinic

·        Disc Suspect (DS)- suggested referral to a glaucoma clinic

·        No Referable Glaucoma (NRG)

·        Ungradable

 For the purposes of our study, both the RG and DS groups will be considered to have “referable glaucoma,” while the NRG group will be maintained as the “no referable glaucoma” group.

 The patient will complete their comprehensive ophthalmology clinic visit after imaging with the camera. The results of the device assessment will be masked, as neither the patient nor the comprehensive ophthalmologist will be informed of the results of the AI-based fundus camera device assessment before or after the clinic visit. Comprehensive ophthalmologists are defined as physicians with at least 3 years of experience after completing ophthalmology residency. All comprehensive ophthalmologists involved in the study currently provide care to patients who visit Aravind Eye Hospital in Pondicherry, India.

 The comprehensive ophthalmologists will follow normal clinic workflow, completing a formal referral to glaucoma clinic if a patient is considered to have “referable glaucoma.” In terms of normal clinic workflow, the comprehensive ophthalmologists will complete a determination for potential glaucoma clinic referral based on the patient’s optic disc appearance, slit lamp exam findings, intraocular pressure(s), a potential gonioscopy and assessment for shallow chambers, family history, and any other eye exam findings or testing they deem relevant. They will be required to document the specific reason(s) for the referral.

 By the end of this stage, participants with “referable glaucoma” will be those referred by:

 Â·        The AI-based fundus camera device only.

·        Both AI-based fundus camera device and the comprehensive ophthalmologist.

·        The comprehensive ophthalmologist only.

 These participants will continue with the testing described in Stage 2 below.

 The remainder of patients will have been determined to have “no referable glaucoma” by both the device and the ophthalmologist. Most of these remaining participants will not undergo any further testing in Stage 2; their involvement in the study will be complete and they will be provided standard of care as per their ocular diagnosis by the comprehensive ophthalmologist. However, for the purposes of specificity testing, a cohort of 60 patients in the “no referable glaucoma” group will be requested to complete the additional tests detailed in Stage 2 below. These patients will be randomly selected as follows: the 10th patient considered to have “no referable glaucoma” will be requested to partake in Stage 2 testing, followed by n+10 until we reach 60 subjects. If a patient refuses, we will defer to the following normal patient (n+11) as necessary.

 Stage 2: Those with “referable glaucoma,” as well as about 60 patients from the “no referable glaucoma” group, will have a 24-2 Humphrey Field Analyzer (HFA, Model 860 HFA 3; Carl Zeiss Meditec Inc, USA) and an image taking of the optic nerve and retinal nerve fiber layer using optical coherence tomography (Model 5000, Cirrus HD-OCT; Carl Zeiss Meditec Inc, USA). They will also be examined by a glaucoma specialist. If the participant cannot stay for testing after their comprehensive ophthalmology clinic visit, the research team will try to schedule a follow-up date for them to complete the HFA and OCT testing.

 The final component of Stage 2 will rely on asynchronous assessments of all "referable glaucoma” participants by a minimum of two glaucoma specialists. Firstly, the first glaucoma specialist will render an initial assessment and diagnosis. This specialist will use all available participant data, HFA, OCT, comprehensive ophthalmology clinic evaluation notes, and relevant patient history from chart review. The specialist will be masked to the results of the Remidio AI as well as the clinical evaluation and findings of comprehensive ophthalmologists. The final diagnosis provided by this specialist will fall into one of the following categories:

  Â·        Definite glaucoma

·        Probable glaucoma (>75% chance of glaucoma)

·        Possible glaucoma (10-75% chance of glaucoma)

·        No glaucoma (<10 % chance of glaucoma)

 Following this initial diagnosis, a second glaucoma specialist will complete the same assessment and diagnosis process as the first specialist with all the available data for each participant. However, he/she will be masked to the determination of the first specialist. Should the second specialist agree with the initial specialist’s diagnosis, the diagnosis will be final. However, should the second specialist disagree with the initial specialist’s categorization, a third glaucoma specialist will adjudicate in this circumstance to make the final diagnosis.

 There will likely be a substantial subset of patients with unreliable HFA and/or OCT testing results due to various reasons including cataracts. In such cases, rather than excluding these patients or completing additional attempts to obtain a reliable test result, the glaucoma specialists will have to incorporate any unreliable testing results into their final glaucoma risk assessment evaluations.