(102 days)
The VIO System is intended to acquire, store, retrieve, display and transfer in vivo images of tissue - including blood vessels, collagen, pigment, stratum corneum, hair shafts or follicles, solar elastosis, atypia, and epidermal disarray - in and through epidermis for review by physicians to assist in forming a clinical judgment. Physicians who interpret VIO System images must have dermatology medical qualifications with skin histology assessment training.
The VIO System (REF V-1000) is a light-based in vivo microscope intended to provide images of skin tissue for review by physicians to assist in forming a clinical judgment. The device does not provide image analysis or diagnostic information. The VIO System uses a handheld reflectance confocal and multiphoton microscope that contacts the skin and optically scans at a depth where most skin conditions originate, without disrupting the tissue. The VIO System delivers low power laser light of a single wavelength (780 nm infrared, Class 1M) into the skin in brief pulses. The resulting signals are acquired while imaging the skin and each signal is assigned to one of four (4) channels: RCM (Reflectance Confocal Microscopy), SHG (Second Harmonic Generation), 2PS (2-Photon Autofluorescence Short), and 2PL (2-Photon Autofluorescence Long). For each image capture, the VIO System combines the four (4) channels to display a single, multi-color image on the VIO System, referred to as "Tetrachrome™." Three (3) view modes are accessible to the user on the VIO System: Standard View, Superficial Zoom, and Deep Zoom. Acquired images are not stored on nor analyzed by the VIO System. The images are saved on a USB flash drive, which is used to transfer images to a commercially available DICOM viewer for interpretation by a physician.
The provided 510(k) summary for the Enspectra VIO System (K232789) outlines the device's acceptance criteria and the clinical study conducted to validate its performance.
Here's a breakdown of the requested information:
1. A table of acceptance criteria and the reported device performance
The summary states: "The effectiveness endpoint for the study was achieved as the results demonstrated overall >90% agreement between each Blinded Reader VIO image assessment and the answer key developed from the Comparative Reader validation assessment." This indicates that the acceptance criterion was >90% agreement.
| Feature/Region | Acceptance Criteria | Reported Blinded Reader Accuracy (95% CI) |
|---|---|---|
| None/Epidermis | >90% Agreement | 96.0% (93.6% to 97.7%) |
| Pigmented cell/Epidermis | >90% Agreement | 93.5% (87.1% to 97.4%) |
| None/Dermis | >90% Agreement | 100% (91.0% to 100%) |
| Collagen/Dermis | >90% Agreement | 98.3% (96.7% to 99.3%) |
| Pigmented cell/Dermis | >90% Agreement | 95.6% (89.0% to 98.8%) |
| Blood vessel/Dermis | >90% Agreement | 87.2% (77.7% to 93.7%) |
| Feature | ||
| Stratum corneum | >90% Agreement | 99.2% (97.1% to 99.9%) |
| Hair shaft or follicle | >90% Agreement | 97.8% (88.2% to 99.9%) |
| Solar elastosis | >90% Agreement | 99.1% (95.1% to 100%) |
| Hyperkeratosis | >90% Agreement | 100% (92.1% to 100%) |
| Atypia | >90% Agreement | 95.6% (84.9% to 99.5%) |
| Epidermal disarray | >90% Agreement | 100% (88.4% to 100%) |
Note: While the overall effectiveness endpoint states >90% agreement, the "Blood vessel/Dermis" accuracy is 87.2% (77.7% to 93.7%). This specific feature's lower bound of the CI falls below 90%, but the report concludes that the "effectiveness endpoint... was achieved as the results demonstrated overall >90% agreement." This suggests that the overall average agreement across all features met the criteria, even if individual feature performance varied.
2. Sample size used for the test set and the data provenance
- Sample Size for Test Set: The study was conducted with N=65 participants. Participants were randomly assigned to either a Training Set or a Testing Set. The exact number of participants solely in the testing set is not explicitly stated as a specific number (e.g., "N=X for test set"). However, it's implied that the "Blinded Readers" were tested on images derived from this N=65 participant pool.
- Data Provenance: The document does not specify the country of origin of the data. The study involved imaging skin lesions on adults (18 years and above) with Fitzpatrick Skin Type I-V, located on the head, neck, limbs, and torso. It is a prospective study as the VIO System was used to image participants during the study.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts
- Number of Experts for Ground Truth: N=3 Comparative Readers were used to develop the answer key (ground truth) for the test set.
- Qualifications of Experts: The Comparative Readers were physicians, and their role involved "validating VIO images against ground truth pathology." The text also states that "Physicians who interpret VIO System images must have dermatology medical qualifications with skin histology assessment training." It's reasonable to infer that the Comparative Readers held these qualifications.
4. Adjudication method (e.g. 2+1, 3+1, none) for the test set
The ground truth was established by "N=3 Comparative Readers developed consensus training materials, then developed an answer key by validating VIO images against ground truth pathology." This implies an expert consensus method among the 3 comparative readers to establish the answer key. The specific adjudication rule (e.g., majority vote, or discussion until agreement) is not detailed, but it's clearly a consensus-based approach rather than simple averaging or individual reading.
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
There is no mention of an MRMC comparative effectiveness study where human readers were compared with and without AI assistance. The study focuses purely on the performance of human readers (Blinded Readers) interpreting VIO images against a ground truth established by other human experts (Comparative Readers). The VIO System itself does not provide image analysis or diagnostic information; it generates images for physician review. Therefore, it's not an AI-assisted interpretation system in the sense of providing a computer-driven initial read or aiding diagnosis.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done
No standalone algorithm performance was done/reported. The device description explicitly states: "The device does not provide image analysis or diagnostic information." The VIO System is an image acquisition device; the interpretation is performed by physicians.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc)
The ground truth for the test set was established by expert consensus of 3 Comparative Readers, validated against pathology. Specifically: "...developed an answer key by validating VIO images against ground truth pathology." This indicates that pathology results served as the definitive reference standard, interpreted and applied to the VIO images by the expert readers.
8. The sample size for the training set
The study involved N=65 participants, randomly assigned to either a Training Set or Testing Set. The exact number of participants specifically in the training set is not explicitly stated as a specific number. However, the N=3 Comparative Readers "developed consensus training materials" which implies a portion of the N=65 dataset was used for this training.
9. How the ground truth for the training set was established
The ground truth for the training set was established by the N=3 Comparative Readers who "developed consensus training materials, then developed an answer key by validating VIO images against ground truth pathology." This suggests a similar process to the test set ground truth, involving expert consensus and correlation with pathology.
§ 878.4580 Surgical lamp.
(a)
Identification. A surgical lamp (including a fixture) is a device intended to be used to provide visible illumination of the surgical field or the patient.(b)
Classification. Class II (special controls). The device, when it is an operating room lamp, a surgical instrument light, a surgical floor standing light, an endoscopic surgical light, a surgical light connector, a ceiling mounted surgical light, a surgical light carrier, surgical light accessories, a surgical lamp, a remote illuminator, or an incandescent surgical lamp, is exempt from the premarket notification procedures in subpart E of part 807 of this chapter subject to the limitations in § 878.9.