Search Results

The Pediatric Vision Scanner is intended for the automated detection of misalignment of the visual axes using polarized light. It is indicated for the screening of amblyopia and microstrabismus associated with amblyopia in children age 2 to 8 undergoing evaluation in a professional eye care setting and who are responsive to taking direction and who can pay attention for at least 5 seconds.

Device Description

The Pediatric Vision Scanner (PVS) is a non-invasive, optical instrument intended to be used for ophthalmic diagnostic purposes. The device integrates all hardware components within the plastic enclosure, including the image acquisition optics, the system computer, and visual display. The PVS is to be held with both hands and level to the ground, with the aperture aimed toward the child. The operator positions the PVS in dim light. Prior to each measurement, a background measurement is obtained to minimize signal interference. Rangefinders (laser diode, 650 nm wavelength) allow the proper positioning of the device (~ 0.5 meters) during background measurement and measurement. After the background is obtained, the operator instructs the patient to open their eyes, peer into the lens (with both eyes), and fixate on a target light (red Light-Emitting Diode (LED)) source (in the shape of a "smiley face"). The operator performs the measurement; a spot of near-infrared (laser diode, 830 nm wavelength), polarized light rotates circularly around the fixation target. Based on the polarization characteristics of the received light, the device uses an algorithm to determine whether or not this light is incident on the nerve fibers emanating from the fovea, the central retinal area for normal visual fixation, of each eye. Based upon this analysis and determination, the following output information is presented to the operator: LED outputs (based on binocularity values), "Tries": total number of scans (performed in two sets of 5, a maximum of 10), "Yield": ratio of number of scans that produced a signal to number of tries, "Binoc" / "binocularity": the ratio of scans resulting in bilateral, simultaneous fixation to number of tries. The device has an optical portion and integrated software analysis to administer, interpret, and convey the results of a scan. The optical portion of the device may be divided into two major pathways: Illumination pathway and Viewing pathway.

AI/ML Overview

Here's a breakdown of the acceptance criteria and the study proving the device meets those criteria, based on the provided text:

Acceptance Criteria and Device Performance

Acceptance Criteria	Reported Device Performance
Clinical Performance:
Sensitivity ≥ 80%	88.81% (95% CI: 82.47 – 93.47%)
Specificity ≥ 82%	91.74% (95% CI: 84.90 – 96.16%)
Non-clinical Performance (Bench - Lower Limit of Detection):
"Pass" results (≥60% binocularity) when prisms inducing deviation below lower detection limit are used.	All devices showed 100% BIN for lower powered prisms, indicating a "Pass." One device showed 60-100% BIN at the detection limit (average 92%), still a "Pass." (Detection limit is between b(4) prism diopters)
"Refer" results (<60% binocularity) when prisms inducing deviation above lower detection limit are in place.	All devices showed 0% BIN for higher powered prisms, indicating a "Refer." (Detection limit is between b(4) prism diopters)
Non-clinical Performance (Bench - Precision):
1. Consistent "pass" results (≥60% binocularity) when no prism (bifoveal fixation).	100% BIN for bifoveal fusion condition across all devices and operators.
2. "Refer" results (<60% binocularity) when prisms inducing a clinically significant deviation are in place.	0% BIN for clinically significant angle (b(4)) across all devices and operators.
3. "Refer" results (<60% binocularity) when prisms inducing a deviation above the lower detection limit are in place.	0% BIN for lower detection limit (b(4)) across all devices and operators.
4. Results independent of device and operator with 90% consistency.	No variability attributed to operator or device at each testing condition; all results consistently met criteria across 10 trials per cell.

Study Information

Sample size used for the test set and the data provenance:
- Sample Size: 252 children were included in the final analysis.
- Data Provenance: Prospective, multi-center, observational investigation conducted at three clinical sites in the United States. These sites were pediatric ophthalmology practices (one clinic at a teaching hospital and one private practice) and one comprehensive ophthalmology clinic.
Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- The ground truth was established by "clinical teams" described as "board-certified pediatric ophthalmologists, orthoptists, optometrists, and ophthalmic technicians." The exact number of individual experts is not specified.
Adjudication method for the test set:
- The text does not explicitly describe an adjudication method beyond stating that the "reference examination" was performed by "clinical teams." It implies that the comprehensive examination itself serves as the definitive ground truth, suggesting a single expert consensus or a standard of care approach rather than a specific adjudication process (e.g., 2+1 or 3+1).
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 comparative effectiveness study involving human readers with and without AI assistance was not done. This study solely evaluated the standalone diagnostic accuracy of the Pediatric Vision Scanner (PVS) against a comprehensive eye examination. The PVS is an automated detection device, not an AI assistance tool for human readers in the context of this report.
If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:
- Yes, a standalone performance study was conducted. The Pediatric Vision Scanner is described as an "optical instrument intended to be used for ophthalmic diagnostic purposes" that uses an "algorithm to determine whether or not this light is incident on the nerve fibers emanating from the fovea." The device provides an automated "Pass" or "Refer" output based on its internal analysis, effectively acting as an algorithm-only diagnostic.
The type of ground truth used:
- Expert Consensus / Standard of Care: The ground truth was a "comprehensive eye examination" (referred to as "reference examination") performed by clinical teams including board-certified pediatric ophthalmologists, orthoptists, optometrists, and ophthalmic technicians. This comprehensive examination included:
  - Visual acuity evaluation
  - Anterior segment evaluation
  - Evaluation of external ocular structures and ptosis
  - Extraocular motility evaluation (including prism-and-cover test or Krimsky test)
  - Evaluation of stereopsis
  - Cycloplegic refraction with retinoscopy
  - Dilated fundus examination
The sample size for the training set:
- The document does not specify a training set sample size. The description pertains to the pivotal clinical trial which evaluated the device's diagnostic accuracy, suggesting it served as the validation/test set for the already developed algorithm.
How the ground truth for the training set was established:
- As no specific training set is mentioned or its data source described in the document, there's no information on how its ground truth was established. The clinical trial described is for performance verification, suggesting the device's algorithm was already developed prior to this pivotal study.

Ask a Question

Ask a specific question about this device

Page 1 of 1