Search Results

The Topcon DRI OCT Triton is a non-contact, high resolution tomographic and biomicroscopic imaging device that incorporates a digital camera for photographing, displaying and storing the retina and surrounding parts of the eye to be examined under Mydriatic and non-Mydriatic conditions.

The DRI OCT Triton 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.

It also includes a Reference Database for posterior ocular measurements which provide for the quantitative comparison of retinal nerve fiber layer, optic nerve head, and the human retina to a database of known normal subjects. The DRI OCT Triton is indicated for use as a diagnostic device to aid in the diagnosis, documentation and management of ocular health and diseases in the adult population.

The DRI OCT Triton ("Triton") and the DRI OCT Triton (plus)") are non-contact, highresolution, tomographic and bio-microscopic imaging systems that merge optical coherence tomography (OCT) and fundus camera into a single device. Triton and Triton (plus) employ the swept source OCT (SS-OCT) technology. Both can take anterior OCT images in addition to fundus OCT images. The fundus camera, in both Triton and Triton (plus), includes color imaging, red-free imaging, and infrared light imaging (hereinafter, IR imaging) capabilities for fundus observation. The Triton (plus) has fluorescein angiography (FA), and fundus autofluorescence angiography (FAF) imaging function in addition to all fundus functions for Triton.

The fundus photographs and OCT images are captured by different system components of this device, which enables Triton to capture an OCT image and a fundus image sequentially. It allows 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. It also has a reference database for posterior ocular measurements of normal subjects, which provide for the quantitative comparison of retinal nerve fiber layer, optic nerve head and the macula.

Captured images are transferred from the device to an off-the-shelf personal computer (PC) via LAN cable, where the dedicated software for this device is installed. The transferred data is then automatically processed with analysis functions such as the automatic retinal layers segmentation, the automatic thickness calculation with several grids, the optic disc analysis and comparison with a reference database of eyes free of ocular pathology, and is finally automatically saved to the PC. It allows the user to manually adjust the automated retinal layer segmentation results and optic disc analysis results.

Accessories include the power cord, chin-rest paper sheet, monitor cleaner, LAN cable; chin-rest paper pins, external fixation target, dust cover accessory case, user manual, unpacking and analysis software DVD.

The Topcon DRI OCT Triton is a non-contact, high-resolution tomographic and biomicroscopic imaging device that incorporates a digital camera for photographing, displaying, and storing data of the retina and surrounding parts of the eye. It is indicated for in vivo viewing, axial cross-sectional, and three-dimensional imaging and measurement of posterior ocular structures (retinal nerve fiber layer, macula, and optic disc) and anterior ocular structures. It includes a Reference Database for posterior ocular measurements to quantitatively compare these structures to a database of known normal subjects. The device is intended as a diagnostic aid in the diagnosis, documentation, and management of adult ocular health and diseases.

1. Table of Acceptance Criteria and Reported Device Performance

The acceptance criteria for the Topcon DRI OCT Triton were based on demonstrating substantial equivalence to its predicate devices, the Topcon 3D OCT-1 Maestro and the Topcon TRC-50DX Retinal Camera. This was evaluated through agreement and precision studies, as well as image quality evaluations. The specific acceptance criteria are implicit in the reported performance metrics shown below.

2. Sample Size Used for the Test Set and Data Provenance

For the Agreement and Precision Study (comparing Triton to Maestro):

• Sample Size: 76 participants, including:
  • 25 Normal eyes
  • 26 Retinal eyes
  • 25 Glaucoma eyes
• Data Provenance: Prospective comparative clinical study conducted at a single U.S. clinical site. (Page 6)

For the Fundus Autofluorescence and Fluorescein Angiography Image Quality Evaluation Study (comparing Triton (plus) to TRC-50DX):

• Specific sample size (number of participants/eyes) is not explicitly stated, but the submission mentions "Majority of the FAF and FA images were graded by both graders" and "response rates (i.e., percentage of subjects whose Triton grades were equal to or better than the corresponding TRC-50DX grades)".
• Data Provenance: Prospective clinical study conducted at one clinical site, located in the United States. (Page 55-56)

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications

• Agreement and Precision Study (Fundus Photographs, Anterior B Scan, Posterior B Scan evaluations): The image quality of the fundus photographs, and the image quality of anterior and posterior OCT B scans were graded by two masked independent experts. (Page 7). Their specific qualifications (e.g., years of experience, specialty) were not detailed in the provided text.
• Fundus Autofluorescence and Fluorescein Angiography Image Quality Evaluation Study: The FAF and FA images were graded by two masked independent graders in a blinded and randomized fashion. Their specific qualifications were not detailed. (Page 56).

4. Adjudication Method for the Test Set

• Agreement and Precision Study: The text describes that two masked independent experts graded the images. For fundus photographs, the company performed a further analysis on image quality grades that differed by 1 point, stating this difference is not considered significantly different when both graders' scores are certain values. For posterior B scans, it states that 34% of images differed by one grade, but all these images were considered clinically useful by both graders. This suggests a form of implicit agreement or tolerance for minor discrepancies, rather than a formal adjudication process like 2+1 or 3+1 where a third expert decides.
• Fundus Autofluorescence and Fluorescein Angiography Image Quality Evaluation Study: The text indicates images were graded by "two masked independent graders" who performed the grading in a "blinded and randomized fashion." No explicit adjudication method (e.g., tie-breaking by a third reader) is mentioned for instances of disagreement.

5. If a Multi Reader Multi Case (MRMC) Comparative Effectiveness Study was done

• A MRMC comparative effectiveness study was not explicitly described for evaluating human reader improvement with AI assistance. The clinical studies focused on comparing the performance and agreement of the Triton device (algorithm included) with predicate devices directly, and on establishing a reference database. Human readers were involved in grading image quality for agreement studies, but the studies were not designed to measure the effect of AI assistance on human reader performance.

6. If a Standalone (i.e. algorithm only without human-in-the loop performance) was done

• The study primarily focuses on the standalone performance of the DRI OCT Triton device compared to predicate devices for quantitative measurements and image quality. The device itself performs the image acquisition, segmentation, and thickness calculations. The evaluation of image quality by experts is a pragmatic assessment of the output generated by the device, essentially evaluating its standalone output for clinical utility. The quantitative metrics (e.g., retinal thickness measurements, optic disc parameters) are direct outputs of the device's algorithms.

7. The Type of Ground Truth Used

• For the Agreement and Precision Study: The ground truth for quantitative measurements was established by comparison against the measurements obtained from the predicate device (Maestro). For image quality evaluations, the ground truth was established by expert grading/consensus from two masked independent experts.
• For the Fundus Autofluorescence and Fluorescein Angiography Image Quality Evaluation Study: The ground truth for image quality was established by expert grading/consensus from two masked independent graders comparing Triton (plus) images to TRC-50DX images.

8. The Sample Size for the Training Set

The provided text describes clinical studies for performance evaluation and for establishing a reference database. It does not explicitly mention a training set sample size for the device's algorithms. The clinical studies appear to be validation studies rather than studies for training the underlying algorithms. The "Reference Database" was established with 410 evaluable eyes. While this database is used for quantitative comparison within the device, it's not explicitly stated to be the training set for the segmentation or measurement algorithms.

9. How the Ground Truth for the Training Set Was Established

Since a "training set" is not explicitly detailed or a method for establishing its ground truth described, this information cannot be provided from the given text. The reference database for quantitative comparisons was established from measurements of 410 normal eyes (age ≥18, no glaucomatous optic nerve damage) collected across six U.S. clinical sites. For these eyes, various scan parameters (full retinal thickness, RNFL thickness, GCL+IPL thickness, GCC thickness, optic disc measurements, TSNIT circle profile measurements) were collected, and percentiles were estimated using quantile regression with age and/or disc area as covariates. This database serves as a "ground truth" for comparison for normal subjects within the device's functionality, but not explicitly as a ground truth for training the segmentation or measurement algorithms themselves.

Ask a specific question about this device