(580 days)
DySIS with Pseudo-Color Imaging (PCI) is a digital colposcope designed to image the cervix and lower genital tract under illumination and magnification. Colposcopy is indicated for women with an abnormal Pap smear in order to affirm normality or detect abnormal appearances consistent with neoplasia, often with directed biopsy. The PCI feature is an adjunctive tool for displaying areas of acetowhitening. It is a tool that should NOT be used as a substitute for a thorough colposcopic evaluation.
DySIS offers non-patient contact, fully digital and high-resolution imaging of the cervix. The field of view is illuminated by an LED; a CCD camera coupled with the imaging optics provides crisp magnified color images. The images can be viewed on a color monitor and on the color touch screen that provides the means to control the device operation. DySIS offers digital tools to fully document the colposcopic examination: Capturing of images for review, storage and printing, video recording. Advanced magnification and imaging. Operator's annotation of suspicious sites. Color-coded mapping of acetowhitening intensity and duration. Database that allows storing and retrieving identification and examination data.
The provided 510(k) summary for the DySIS Digital Colposcope details non-clinical tests to support substantial equivalence but explicitly states that no clinical studies were conducted to demonstrate improved clinical performance using the PCI feature sufficient to claim superiority to conventional colposcopy. The device's PCI feature is described as an adjunctive tool which should not be used as a substitute for a thorough colposcopic evaluation.
Therefore, while the summary describes some non-clinical performance specifications, it does not include acceptance criteria for clinical performance that the device meets, nor does it describe a study proving such clinical performance.
However, it does provide information about non-clinical tests that were performed.
Here's a breakdown of the requested information based on the provided text, with the caveat that clinical performance acceptance criteria and proof are absent:
Acceptance Criteria and Device Performance (Non-Clinical)
| Acceptance Criteria (Non-Clinical) | Reported Device Performance |
|---|---|
| LED Temperature: No significant increase after 1 hour continuous exposure | Paper did not experience any significant increase in temperature after one hour of continuous exposure. |
| Timing Tolerance (Imaging Delay): Small and not statistically significant differences from AUC values where no delay occurred. | Differences from AUC values where no delay occurred were small and not statistically significant. |
| Timing Tolerance (Acetic Acid Delay): Small and not statistically significant differences from AUC values where no delay occurred. | Differences from AUC values where no delay occurred were small and not statistically significant. |
| Acetic Acid Concentration: Accurately depict acetowhitening effect with both 3% and 5% acetic acid solutions. | The DySIS map is capable of accurately depicting the aceto-whitening effect regardless of whether a 3% or 5% acetic acid solution is used. (Note: 5% showed a more extreme effect, as expected) |
| Color Calibration: Pixel Value fell within the desired tolerance band of ±2 from points within a random distance between +3 mm from focusing distance. | Pixel Value fell within the desired tolerance band of ±2 from points within this range. |
| Image Alignment: Compensation to within acceptable level for distortions equivalent to 7 mm of actual movement. | Algorithm was able to compensate to 0.4 mm, well within the acceptable level, for distortions totaling 20% change from original (equivalent to 7 mm movement). |
Detailed Study Information
Given the nature of the submission (510(k) for substantial equivalence based on non-clinical tests and adjunctive feature), much of the requested clinical study information is explicitly stated as not applicable or not performed for establishing clinical performance.
-
Sample sizes used for the test set and the data provenance:
- Clinical Test Set: Not applicable for establishing clinical performance superiority or non-inferiority. The "test results using patient data" for "image registration, acetic acid timing errors, and range of PCI values" were described as retrospective patient data. The sample size for this retrospective data is not specified.
- Non-Clinical Tests:
- LED Temperature: Not applicable (tested on a piece of white paper).
- Timing Tolerance: Not specified (refers to "AUC values").
- Acetic Acid Concentration: Not specified (refers to "the 5% solution," suggesting a comparison, but no sample size).
- Color Calibration: Not specified ("An operator was asked to focus...").
- Image Alignment: One cervical image distorted ten times.
-
Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- Not applicable as no clinical study to establish improved clinical performance was conducted. The ground truth for the retrospective patient data used for "image registration, acetic acid timing errors, and range of PCI values" is not explicitly stated as being established by experts in this context.
-
Adjudication method (e.g. 2+1, 3+1, none) for the test set:
- Not applicable as no clinical study to establish improved clinical performance was conducted.
-
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 done to demonstrate that the PCI feature is superior to conventional colposcopy or that human readers improve with AI (PCI) assistance. The text explicitly states: "No clinical studies were conducted to demonstrate improved clinical performance using the PCI feature," and "Importantly, the study using retrospective patient data did not demonstrate that the PCI feature is superior to conventional colposcopy in helping the clinician target sites with suspected CIN for biopsy."
-
If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:
- Yes, in the sense that the PCI feature generates a "color-coded mapping of acetowhitening" based on image processing. This is an algorithm-only output. However, the use of this output is adjunctive and not a substitute for the human clinician's thorough colposcopic evaluation. The device is designed so the clinician must perform standard colposcopy and identify biopsy sites before viewing the PCI image.
-
The type of ground truth used (expert consensus, pathology, outcomes data, etc.):
- For the non-clinical tests, the "ground truth" was based on controlled experimental setups and measurements (e.g., thermal output, pixel values, measured distortions).
- For the retrospective patient data used to evaluate basic performance (image registration, timing errors, PCI range), the nature of the ground truth is not specified but likely refers to the original clinical assessments or records from which the data was drawn. It was not used to establish a clinical ground truth against which the PCI's diagnostic accuracy was measured.
-
The sample size for the training set:
- Not applicable / Not specified. The document does not describe a machine learning model that underwent a distinct training phase. While the PCI feature involves "processing digital images," it's presented as an image processing technique rather than a trainable AI model in the modern sense requiring a 'training set.'
-
How the ground truth for the training set was established:
- Not applicable. As no training set for an AI model is described, there's no mention of how its ground truth would have been established.
§ 884.1630 Colposcope.
(a)
Identification. A colposcope is a device designed to permit direct viewing of the tissues of the vagina and cervix by a telescopic system located outside the vagina. It is used to diagnose abnormalities and select areas for biopsy. This generic type of device may include a light source, cables, and component parts.(b)
Classification. Class II (special controls). The device, when it is a standard colposcope (or colpomicroscope) that uses only a white light source, does not use filters other than a green filter, does not include image analysis software, and is not smartphone-based, is exempt from the premarket notification procedures in subpart E of part 807 of this chapter subject to the limitations in § 884.9.