(89 days)
Harmony is a comprehensive software platform intended for use in importing, processing, measurement, analysis and storage of clinical images and videos of the eye as well as in management of patient data, clinical information, reports from ophthalmic diagnostic instruments through either a direct connection with the instruments or through computerized networks.
Harmony is a modification to the existing Harmony cleared in K182376. The differences between the new version and the currently cleared version are modifications to the graphical user interface consisting of PixelSmart Technology, Internationalization support, Analytical thickness grids, Hanging protocols, and Automatic image smoothing while zooming in.
Harmony is a comprehensive software platform intended for use in importing, processing, measurement, analysis and storage of clinical images and videos of the eye, as well as for management of patient data, diagnostic data, clinical information, reports from ophthalmic diagnostic instruments through either a direct connection with the instruments or through computerized networks.
Harmony is used together with a number of computerized digital imaging devices, including:
- Optical Coherence Tomography devices .
- Mydriatic retinal cameras .
- Non-mydriatic retinal cameras .
- Biomicroscopes (slit lamps)
In addition, Harmony collects and manages patient demographics, image data, and clinical reports from a range of medical devices, including:
- Scanning Laser Ophthalmoscope images and videos .
- Non Radiometric Ultrasound devices ●
- Video image sources ●
- TWAIN compliant imaging sources ●
- Compliant data sources placed in network accessible folders and directories
- . Images of known format from digital cameras and scanners
- . Printer files of known format form computerized diagnostic devices
- Electronic information complying to accepted DICOM formats
- Other devices connected in proprietary formats ●
There are 5 notable device modifications subject of this submission: PixelSmart Technology, International support, Analytical thickness grids, Hanging protocols, and Automatic image smoothing while zooming in, along with some minor modifications.
PixelSmart is an optional post-processing image enhancement algorithm performing a moving average across OCT B-scans, reducing speckle noise and improving contrast by applying smoothing.
International support adds the possibility to use the Harmony user interface and online user manual in Spanish, in addition to the standard English software.
Analytical thickness grids offer the same functionality as the existing, cleared thickness grids in Topcon's IMAGEnet 6, now also in Harmony. The grids show sectorial average thickness values as derived from OCT segmentation data.
Hanging protocols allows a customizable image display arrangement in the Harmony user interface, resembling the arrangement of physical images on a light box.
Automatic image smoothing while zooming in is an optional display feature that will cause OCT B-scan images on higher zoom levels to look less pixelated.
The provided text describes a 510(k) premarket notification for a device called "Harmony" by Topcon Healthcare Solutions. This submission is for modifications to an existing cleared device (K182376). As such, the focus is on demonstrating that the modifications do not introduce new safety or effectiveness concerns and that the device remains substantially equivalent to its predicate.
Therefore, the document does not contain the kind of detailed clinical study and performance data (e.g., acceptance criteria tables, sample sizes for test/training sets, expert ground truth establishment, MRMC studies) that would typically be required for the initial clearance of a novel AI/ML-driven device with diagnostic claims. Instead, it relies on demonstrating that the "modified Harmony" functions equivalently to the predicate Harmony, primarily through software validation and verification.
Based on the provided text, here's what can and cannot be answered:
1. A table of acceptance criteria and the reported device performance
- Cannot be provided. The document states: "Software validation and verification demonstrate that Harmony performs as intended and meets its specifications, using methods equivalent to the predicate device." However, it does not specify quantitative acceptance criteria for performance metrics (e.g., sensitivity, specificity, accuracy, F1-score) or report specific performance values for the modified features. This is expected given that the modifications are primarily related to UI, image enhancement (PixelSmart), and display features, not fundamental diagnostic algorithms requiring extensive performance studies against clinical ground truth.
2. Sample sizes used for the test set and the data provenance (e.g. country of origin of the data, retrospective or prospective)
- Cannot be provided in detail. The document mentions "software validation and verification activities" and "non-clinical performance testing." These are typically done with internal test cases or simulated data rather than large, independent clinical test sets for a device of this nature (an image management and processing system with UI/display modifications). There is no mention of specific sample sizes of patient images or their provenance.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g. radiologist with 10 years of experience)
- Not applicable/Cannot be provided. Since no formal clinical test set with a "ground truth" adjudicated by multiple experts is described for the modifications in this 510(k) summary, details about expert involvement are not present. The changes (PixelSmart, Internationalization, Analytical thickness grids, Hanging protocols, Automatic image smoothing) relate to image display, processing, and user interface, rather than directly generating a diagnostic output that would require expert-adjudicated ground truth.
4. Adjudication method (e.g. 2+1, 3+1, none) for the test set
- Not applicable/Cannot be provided. As no multi-expert ground truth establishment for a test set is described, there's no mention of an adjudication method.
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, an MRMC study was not done. The document describes modifications to an image management and processing system. The "PixelSmart" technology is an optional post-processing image enhancement algorithm (moving average to reduce speckle noise and improve contrast). While this could hypothetically improve reader performance, the submission does not present an MRMC study to quantify such an effect. This type of study is more common for AI algorithms directly assisting in interpretation or detection, which is not the primary claim for these modifications.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done
- Not explicitly described as a formal validation study. The "PixelSmart" feature is an algorithm (moving average). Its performance would be evaluated internally for its intended effect (reducing speckle noise, improving contrast). However, the document does not present a standalone performance study with metrics like sensitivity/specificity for a specific clinical task. The assessment is that it "performs as intended" and "meets its specifications" as an image enhancement tool.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc)
- Not applicable/Cannot be provided. The modifications are not addressing a diagnostic claim that would require ground truth from expert consensus, pathology, or outcomes data. The "ground truth" for verifying these changes would relate to software functionality (e.g., does PixelSmart correctly apply a moving average? Does the Spanish UI display correctly?).
8. The sample size for the training set
- Not applicable/Cannot be provided. The "Harmony" system itself is a software platform. While the PixelSmart feature is an algorithm, the document does not describe it as a machine learning model that undergoes a "training" phase with a large dataset. It's described as a "moving average across OCT B-scans," suggesting a rule-based or conventional image processing algorithm rather than a deep learning model. Therefore, there's no mention of a training set size.
9. How the ground truth for the training set was established
- Not applicable/Cannot be provided. As there's no description of a training set, the method for establishing its ground truth is not provided.
Summary of what is described regarding the study/validation:
- Type of Study: Software validation and verification, and non-clinical performance testing.
- Purpose: To demonstrate that the modified Harmony functions equivalently to the predicate Harmony and that the modifications do not introduce new safety or effectiveness concerns.
- Assessment: Risk assessment was conducted, and "newly identified risks or modified existing risks are mitigated, and no unacceptable risk was identified."
- Standards Followed: IEC 62304 (Medical Device Software Life Cycle Processes), NEMA PS 3.1-3.20 (DICOM), ISO IEC 10918-1 (JPEG), ISO 14971 (Risk Management).
In essence, this 510(k) relies on demonstrating the equivalence of a modified, already cleared, non-diagnostic software platform through robust engineering and software validation principles, rather than extensive clinical performance studies common for novel AI diagnostic devices.
§ 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).