(152 days)
The CIRRUS photo is a non-contact, high resolution tomographic and biomicroscopic imaging device that incorporates a digital camera which is suitable for photographing, displaying and storing the data of the retina and surrounding parts of the eye to be examined under mydriatic and non-mydriatic conditions.
These photographs support the diagnosis and subsequent observation of eye diseases which can be visually monitored and photographically documented. The CIRRUS photo is indicated for in vivo viewing, axial cross sectional, and three-dimensional imaging and measurement of posterior ocular structures, including retinal nerve fiber layer, macula and optic disc as well as imaging of anterior ocular structures and measurement of central corneal thickness.
It also includes a Retinal Nerve Fiber Layer (RNFL), Optic Nerve Head (ONH), and Macular Normative Database which are quantitative tools for the comparison of retinal nerve fiber layer, optic nerve head, and the macula in the human retina to a database of known normal subjects. It is intended for use as a diagnostic device to aid in the detection and management of ocular diseases including, but not limited to, macular holes, cystoid macular edema, diabetic retinopathy, age-related macular degeneration, and glaucoma.
The CIRRUS photo is a computerized optical instrument that combines the diagnostic and imaging capabilities of the Carl Zeiss Meditec VISUCAM PRO NM Digital Camera and the Carl Zeiss Meditec Cirrus HD-OCT Optical Coherence Tomographer Model 4000. The CIRRUS photo was developed to provide both subjective and objective imaging, and to optimize space by combining fundus photography and spectral domain optical coherence tomography, allowing the anterior or posterior segments of the eye to be viewed and photographically documented with the pupil in a non-mydriatic state, within the same instrument. To optimize the workflow, the system applies the same optical beam delivery system for imaging and scanning.
The CIRRUS photo consists of a Fundus Camera Main Unit and a spectral domain optical coherence tomographer (SD-OCT) Module, both of which are installed on a single instrument table. The CIRRUS photo is operated via computer mouse, keyboard and joystick as part of the base of the main unit and an external monitor is mounted on top of the instrument table.
The CIRRUS photo is offered in two models. Model 600 and Model 800. Fundus auto fluorescence is available on both the Model 600 and 800; the Model 800 also offers fluorescein angiography and indocyanine green angiography (ICGA).
CIRRUS photo data can be analyzed using the same Cirrus HD-OCT algorithms and normative database cleared under K111157 (Cirrus HD-OCT software version 6.0), including Advanced Retinal Pigment Epithelium (RPE) Analysis, Guided Progression Analysis (GPA) for Optic Nerve Head (ONH) parameters, and Ganglion Cell Analysis, and the Ganglion Cell Normative Database. As these algorithms and database reside separately on the Cirrus HD-OCT version 6.0 software, the analyses of the CIRRUS photo data are carried out using an external Cirrus Review station.
The CIRRUS photo Models 600 and 800 underwent several studies to demonstrate substantial equivalence to predicate devices and to support expanded indications for use. Key studies focused on evaluating the repeatability, reproducibility, and comparability of measurements between the CIRRUS photo and the Cirrus HD-OCT Model 4000 for Central Corneal Thickness (CCT), RPE illumination/elevation, and Ganglion Cell Analysis (GCA) parameters.
The primary method for establishing acceptance criteria and demonstrating performance was through equivalence studies, comparing the new device against a predicate device (Cirrus HD-OCT Model 4000) for key measurement parameters. The acceptance criteria generally involved demonstrating that the mean difference between measurements from the CIRRUS photo and the predicate device was close to zero, with narrow 95% Confidence Intervals and Limits of Agreement, and that repeatability and reproducibility standard deviations were small.
Here's a breakdown of the acceptance criteria and study details:
1. Table of Acceptance Criteria and Reported Device Performance
Acceptance Criteria (Implicit / Demonstrated Equivalence to Predicate, and good Repeatability/Reproducibility)
| Metric/Parameter | Acceptance Criteria (Implicit from Study Design) | Reported Device Performance (CIRRUS photo vs. Cirrus HD-OCT) |
|---|---|---|
| Central Corneal Thickness (CCT) | Mean difference between devices close to 0; 95% CI of mean difference crossing 0 or very small; 95% Limits of Agreement clinically acceptable. Good repeatability and reproducibility (small SD and COV). | Mean Difference (CIRRUS photo - Cirrus 4000): |
- Phase I: -0.89 µm (95% CI: -2.85, 1.06)
- Phase II: 1.20 µm (95% CI: 0.22, 2.19)
Repeatability (SD): 4.49 µm (COV: 0.8%)
Reproducibility (SD): 5.216 µm (COV: 0.96%)
Performance meets implicit criteria for agreement and precision. |
| Area of Sub-RPE Illumination (mm²) | Mean difference between devices close to 0; 95% CI of mean difference crossing 0 or very small; 95% Limits of Agreement clinically acceptable. Good repeatability and reproducibility (small SD and COV). | Mean Difference (CIRRUS photo - Cirrus HD-OCT): - 200x200 Scan: -0.0108 mm² (95% CI: -0.1632, 0.1416)
- 512x128 Scan: -0.0314 mm² (95% CI: -0.1742, 0.1115)
Repeatability (SD): 0.1099 mm² (200x200), 0.1329 mm² (512x128)
Reproducibility (SD): 0.1774 mm² (200x200), 0.2442 mm² (512x128)
Performance meets implicit criteria for agreement and precision. |
| Closest Distance to Fovea (mm) | Mean difference between devices close to 0; 95% CI of mean difference crossing 0 or very small; 95% Limits of Agreement clinically acceptable. Good repeatability and reproducibility (small SD and COV). | Mean Difference (CIRRUS photo - Cirrus HD-OCT): - 200x200 Scan: 0.01 mm (95% CI: -0.01, 0.02)
- 512x128 Scan: 0.02 mm (95% CI: -0.00, 0.03)
Repeatability (SD): 0.0343 mm (200x200), 0.0397 mm (512x128)
Reproducibility (SD): 0.0447 mm (200x200), 0.0571 mm (512x128)
Performance meets implicit criteria for agreement and precision. |
| Area of RPE Elevation (mm²) | Mean difference between devices close to 0; 95% CI of mean difference crossing 0 or very small; 95% Limits of Agreement clinically acceptable. Good repeatability and reproducibility (small SD and COV). | Mean Difference (CIRRUS photo - Cirrus HD-OCT, 3mm Circle): - 200x200 Scan: -0.040 mm² (95% CI: -0.098, 0.018)
- 512x128 Scan: -0.004 mm² (95% CI: -0.074, 0.066)
Reproducibility COV%: 8.30% (200x200, 3mm Circle), 8.89% (512x128, 3mm Circle)
Performance meets implicit criteria for agreement and precision. |
| Volume of RPE Elevation (mm³) | Mean difference between devices close to 0; 95% CI of mean difference crossing 0 or very small; 95% Limits of Agreement clinically acceptable. Good repeatability and reproducibility (small SD and COV). | Mean Difference (CIRRUS photo - Cirrus HD-OCT, 3mm Circle): - 200x200 Scan: -0.0024 mm³ (95% CI: -0.0050, 0.0002)
- 512x128 Scan: 0.0002 mm³ (95% CI: -0.0029, 0.0034)
Reproducibility COV%: 8.45% (200x200, 3mm Circle), 9.07% (512x128, 3mm Circle)
Performance meets implicit criteria for agreement and precision. |
| Ganglion Cell Analysis (GCA) Parameters (e.g., Average GCL+IPL Thickness) | Mean difference between devices close to 0; 95% CI of mean difference crossing 0 or very small; 95% Limits of Agreement clinically acceptable. Good repeatability and reproducibility (small SD and COV). | Mean Difference (CIRRUS photo - Cirrus HD-OCT, Average Thickness): - Normal Eyes: -0.3 µm (95% CI: -0.5, -0.2)
- Diseased Eyes: -0.3 µm (95% CI: -0.5, -0.2)
Reproducibility COV% (Average Thickness): 0.97% (Normal), 0.98% (Diseased)
Performance meets implicit criteria for agreement and precision across numerous GCA parameters. |
| GCA Normative Database Transference | CIRRUS photo measurements should be comparable to Cirrus HD-OCT to allow adjustment and use of the established normative database. | Regression analysis was used to adjust the Cirrus HD-OCT GCA Normative Database for use with CIRRUS photo, creating an "adjusted CIRRUS photo Normative Database." This indicates successful transference and adjustment. |
2. Sample size used for the test set and the data provenance (e.g. country of origin of the data, retrospective or prospective)
The studies were prospective. The country of origin is not explicitly stated, but the company is based in Germany with a US contact, suggesting a likely multi-site international or US-based study given the English language requirement in one study.
Test Set Sample Sizes:
- Central Corneal Thickness (CCT) Study:
- Phase I (inter-device variability): 29 subjects enrolled, 28 included in analysis.
- Phase II (inter-operator variability): 23 subjects enrolled and qualified for analysis.
- Area of Increased Illumination under RPE (dry AMD with GA) Study: 21 enrolled subjects, with at least one eye qualifying. Mean age 79.8 years.
- Elevated RPE (dry AMD with drusen) Study: 31 enrolled subjects, with at least one eye qualifying. Mean age 80.1 years.
- Ganglion Cell Analysis (GCA) Normal Eyes Study:
- Phase I (inter-operator variability): 30 subjects.
- Phase II (inter-device variability): 33 subjects. Combined mean age 43.5 years.
- Ganglion Cell Analysis (GCA) Diseased Eyes (Glaucoma) Study: 77 subjects enrolled, 68 included in analysis (37 mild, 16 moderate, 13 severe, 2 end-stage glaucoma). Mean age 67.4 years.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts
The studies described are primarily comparability, repeatability, and reproducibility studies, not diagnostic accuracy studies against a definitive ground truth established by experts in the traditional sense. The "ground truth" here is the measurement obtained by the predicate device (Cirrus HD-OCT Model 4000) and the consistency of measurements within and between devices/operators.
For the RPE analyses (increased illumination and elevated RPE), operators reviewed and manually edited the automated segmentation algorithm output when necessary, consistent with instructions. This suggests that the operators (who were administering the tests) were also implicitly acting as "experts" for reviewing and refining the data derived by the algorithm, effectively enhancing the quality of the algorithmic output, but not establishing a separate, independent ground truth. The number and qualifications of these "operators" are not explicitly stated as "experts" in the clinical evaluation sense, but rather as trained personnel for device operation and data review.
4. Adjudication method (e.g. 2+1, 3+1, none) for the test set
There was no explicit independent adjudication method (like 2+1 or 3+1) described for establishing ground truth from multiple experts. For the RPE studies, the "ground truth" was derived from the automated segmentation algorithms, with manual editing by the operating personnel, but without a formal multi-reader consensus process indicated. The focus was on comparing measurements between devices and assessing measurement consistency.
5. If a multi reader multi case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance
No MRMC comparative effectiveness study was conducted or described in the provided text. The studies focused on the performance of the device's measurement algorithms themselves (standalone performance and comparability to a predicate device), not on how human readers' diagnostic accuracy is improved with or without AI assistance from this specific device.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done**
Yes, standalone performance was implicitly evaluated through the comparability, repeatability, and reproducibility studies. The device measures various parameters (CCT, RPE illumination/elevation, GCA thickness values) using its inherent algorithms. The comparison against the predicate device (Cirrus HD-OCT Model 4000) essentially evaluates the standalone measurement capability of the CIRRUS photo's algorithms against an established standard. While human operators are involved in operating the device and in some cases, reviewing and correcting automated segmentations (especially for RPE analyses), the core "performance" being evaluated is the accuracy and precision of the numerical measurements generated by the device's algorithms.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.)
The "ground truth" in these studies was primarily comparative to a predicate device (Cirrus HD-OCT Model 4000) and statistical measures of repeatability and reproducibility. For the advanced RPE analysis, the ground truth for the segmentation of increased illumination areas and RPE elevations was the output of the device's automated segmentation algorithm, refined by manual editing from the operators. There is no mention of pathology, expert consensus (beyond operator review), or long-term outcomes data as a definitive ground truth for individual measurements.
8. The sample size for the training set
The training set information is only explicitly provided for the Ganglion Cell Normative Database, which was adapted for the CIRRUS photo.
- GCA Normative Database: Derived from an additional analysis of the Cirrus HD-OCT Macular Thickness normative database, which included 282 subjects. These subjects were aged 19-84 years, considered representative of a normal population, and data was collected from seven sites.
For the other advanced algorithms (e.g., Advanced RPE Analysis, GPA), the document states they "reside separately on the Cirrus HD-OCT version 6.0 software" and are carried out using an external Cirrus Review station. This implies these algorithms were developed and likely trained prior to this submission, on data associated with the predicate Cirrus HD-OCT system (K111157), but specific training set sizes for these individual algorithms are not provided within this document.
9. How the ground truth for the training set was established
For the Ganglion Cell Normative Database (and by extension the Macular Thickness normative database):
- The ground truth was established by identifying subjects deemed "representative of a normal population."
- Scans were acquired from these normal subjects.
- Segmentation algorithms were applied to identify the thickness of specific layers (e.g., combined ganglion cell plus inner plexiform layers for GCA).
- These measurements from normal subjects then formed the reference values for the normative database, which is age-corrected. This database essentially defines "normal" based on statistical distribution from this large sample.
§ 886.1570 Ophthalmoscope.
(a)
Identification. An ophthalmoscope is an AC-powered or battery-powered device containing illumination and viewing optics intended to examine the media (cornea, aqueous, lens, and vitreous) and the retina of the eye.(b)
Classification. Class II (special controls). The device, when it is an AC-powered opthalmoscope, a battery-powered opthalmoscope, or a hand-held ophthalmoscope replacement battery, is exempt from the premarket notification procedures in subpart E of part 807 of this chapter subject to the limitations in § 886.9.