FDA 510(k) Search - By Innolitics (SaMD and AI Experts)

The Retina Workplace is a FORUM® application intended for processing and displaying image and optical coherence tomography data. It is also intended for generating reports that contain results from optical coherence tomography and fundus photography.

The Retina Workplace uses CIRRUS™ algorithms and normative databases as quantitative tool for the comparison of macular thickness data to a database of normal subjects.

The Retina Workplace is intended to aid trained healthcare professionals in the detection, monitoring and management of ocular diseases including, but not limited to, macular holes, cystoid macular edema, diabetic retinopathy and age-related macular degeneration.

Device Description

The Retina Workplace is a software application to a Class II medical device regulated under 21 CFR §892.2050 and therefore is a Class II medical device. As a software application of the FORUM PACs system (server), the Retina Workplace client is designed to process and display CIRRUSTM HD optical coherence tomography (OCT) data and images provided by Fundus photography.

More specifically, the Retina Workplace processes the retrieved OCT exam data from CIRRUS HD-OCT and CIRRUS Photo images from the FORUM server to display the results of macular thickness analysis (MTA), macular change analysis (MCA), and 5 line raster analysis. The results may be compared against an age-matched population of normal subjects. Analysis and databases comparison is performed by using algorithms and normative databases previously cleared in CIRRUS HD-OCT (K111157) and CIRRUS Photo (K112184) predicate devices. The reports generated by the Retina Workplace can be displayed, printed, and the patient record can be stored back on the FORUM server database.

In addition, Retina Workplace can automatically register images generated by Fundus photography with the OCT cube image when provided in a valid DICOM format, including the necessary pixel spacing. Additionally, the user has the option to register the fundus / OCT images manually. The automatic/manual image registration features (Fovea Finder™ and AutoCenter™) are currently cleared on the CIRRUS™ HD-OCT system (K111157) and CIRRUSTM Photo (K112184) devices.

The Retina Workplace provides the same overlays as included in the currently cleared CIRRUS HD-OCT system such as:

Fundus image overlay with another fundus image. (Fundus to Fundus registration)
ILM-RPE overlay on the fundus image. (Inner limiting membrane / retinal pigment epithelium)
ETDRS overlay on the fundus image. (Early Treatment of Diabetic Retinopathy Study)

AI/ML Overview

The provided text describes the Retina Workplace, a software application that processes and displays image and optical coherence tomography data, and generates reports. It also indicates that the device has undergone verification and validation testing to ensure it meets acceptance criteria.

However, the document does not contain a detailed table of specific acceptance criteria or quantitative performance metrics for the device itself. Instead, it focuses on demonstrating substantial equivalence to predicate devices and adherence to internal software development and risk management processes.

Therefore, many of the requested details, particularly quantitative performance metrics and specific study designs for proving those metrics, are not present in the provided text.

Here's a breakdown of the information that can be extracted and the information that is not available:

1. Table of Acceptance Criteria and Reported Device Performance

This specific table, with quantitative acceptance criteria and corresponding performance metrics, is not explicitly provided in the document. The document states that "software verification and regression testing has been performed successfully to meet their previously determined acceptance criteria" and that "Verification and validation activities met their acceptance criteria successfully and proved that the product, the Retina Workplace, meets its requirements and intended uses." However, the exact thresholds and measured performance values are not detailed.

General statements on performance found in the document:

"Verification and validation testing was performed to evaluate the performance and functionality of the software and the device has been found to perform as intended."
"In a direct comparison test, the original OCT test reports generated on the CIRRUS HD-OCT and CIRRUS photo were compared to the reports generated by the Retina Workplace using the same test data, to verify that the results contained in both reports were equivalent." (Implies an acceptance criterion of "equivalence" in report generation.)
The device "does not introduce any new potential safety risks and is substantially equivalent to, and performs as well as, the predicate and reference devices."

Regarding the study that proves the device meets acceptance criteria:

The study referenced is primarily a non-clinical validation test and software bench testing.

2. Sample size used for the test set and the data provenance (e.g. country of origin of the data, retrospective or prospective)

Test set sample size: "representative data (sample data that is representative of clinical cases)" was used, but the exact sample size (number of cases or images) is not specified.
Data provenance: "sample data that is representative of clinical cases" is mentioned, but the country of origin of the data or whether it was retrospective or prospective is not specified.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts

Number of experts: "Validation of clinical functionalities (Use Case) was completed by ophthalmologists". The exact number of ophthalmologists is not specified.
Qualifications of experts: The experts are identified as "ophthalmologists". Specific details such as "10 years of experience" are not provided.

4. Adjudication method (e.g. 2+1, 3+1, none) for the test set

The document states that ophthalmologists "executed test cases that simulated the use of the device in a clinical environment and completed questionnaires rating the various aspects of the software." This suggests their feedback was used to validate functionality. However, a formal adjudication method like "2+1" or "3+1" for reaching a consensus on ground truth or performance assessment is not described. It appears to be an assessment by the ophthalmologists, but how disagreements or variations in their ratings were handled is not detailed.

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 involving AI assistance for human readers is described. The study focused on validating the software's functionality and equivalence to existing cleared functionalities. The Retina Workplace itself is positioned as an accessory to a PACS system, providing tools and algorithms (CIRRUS algorithms and normative databases, which were previously cleared). The document does not describe studies where human readers are observed using the Retina Workplace to see an improvement in their diagnostic performance. Therefore, an effect size of human improvement with AI assistance is not provided.

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

The Retina Workplace is described as a "software application" that "processes and displays image and optical coherence tomography data" and "uses CIRRUS™ algorithms and normative databases as quantitative tool for the comparison of macular thickness data to a database of normal subjects."
"In a direct comparison test, the original OCT test reports generated on the CIRRUS HD-OCT and CIRRUS photo were compared to the reports generated by the Retina Workplace using the same test data, to verify that the results contained in both reports were equivalent." This suggests a standalone evaluation of its report generation function against established methods.
The Retina Workplace "processes the retrieved OCT exam data from CIRRUS HD-OCT and CIRRUS Photo images from the FORUM server to display the results of macular thickness analysis (MTA), macular change analysis (MCA), and 5 line raster analysis." It specifically states that "The MTA, MCA, and 5 line raster analysis software applications and normative databases remain unchanged from the currently cleared versions in the CIRRUS HD-OCT and CIRRUS Photo products." This implies that the core algorithms' standalone performance was previously established during the clearance of the predicate CIRRUS devices and is not re-evaluated for this submission, as they are being reused without change.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc)

For the "direct comparison test" of report generation, the "original OCT test reports generated on the CIRRUS HD-OCT and CIRRUS photo" served as the reference or ground truth for equivalence. These predicate devices' reports are assumed to be "true" or at least the standard for output.
For the validation of clinical functionalities by ophthalmologists, their assessment based on their expertise ("completed questionnaires rating the various aspects of the software") served as a form of expert assessment/consensus on the software's usability and fulfillment of clinical needs.

8. The sample size for the training set

The document describes the Retina Workplace as using "CIRRUS™ algorithms and normative databases". It explicitly states, "The MTA, MCA, and 5 line raster analysis software applications and normative databases remain unchanged from the currently cleared versions in the CIRRUS HD-OCT and CIRRUS Photo products." This means the normative databases and algorithms were trained and established previously during the clearance of the predicate devices (K111157 and K112184). The training set and its size for those original algorithms/databases are not provided in this document (K150467). This submission focuses on the integration and display of results from these pre-existing algorithms.

9. How the ground truth for the training set was established

As mentioned above, the core algorithms and normative databases pre-exist from predicate devices. How their ground truth was established for their original training is not detailed in this K150467 summary. It's implied that this was handled in the previous clearances (K111157 and K112184).

Summary

Regulation Number and Section

§ 892.2050 Medical image management and processing system.

(a)
Identification. A medical image management and processing system is a device that provides one or more capabilities relating to the review and digital processing of medical images for the purposes of interpretation by a trained practitioner of disease detection, diagnosis, or patient management. The software components may provide advanced or complex image processing functions for image manipulation, enhancement, or quantification that are intended for use in the interpretation and analysis of medical images. Advanced image manipulation functions may include image segmentation, multimodality image registration, or 3D visualization. Complex quantitative functions may include semi-automated measurements or time-series measurements.(b)
Classification. Class II (special controls; voluntary standards—Digital Imaging and Communications in Medicine (DICOM) Std., Joint Photographic Experts Group (JPEG) Std., Society of Motion Picture and Television Engineers (SMPTE) Test Pattern).