Search Results

This software is an accessory to standard colonoscopy. It is intended for use in the viewing, recording, archival, localization, documentation, and retrieval of still images, video, and patient data during and after a standard colonoscopic procedure. Standard colonoscopy is indicated for the evaluation of results from an abnormality on barium enema or other imaging study, unexplained gastrointestinal bleeding, screening and surveillance for colonic neoplasia, the excision of a colonic polyp, or the management of chronic inflammatory bowel disease.

Captured, compressed videos from previous exams are for viewing and reference purposes and are not intended for primary diagnosis.

Colonoscopy Assistant is a software application designed to provide a streamlined clinical user interface for colonoscopy. The software serves as a portal to useful information before, during, and after a colonoscopic exam. The software displays live video from the colonoscope, enables high-resolution image capture, provides digital noise reduction, displays side-by-side playback of previous exams, estimates the scope camera location during a colon video, and stores all patient and exam information.

This submission (K102949) describes the Colonoscopy Assistant, a software application designed to streamline the clinical user interface for colonoscopy by providing tools for viewing, recording, archiving, localization, documentation, and retrieval of still images, video, and patient data.

Acceptance Criteria and Reported Device Performance

The submission does not explicitly state quantitative acceptance criteria or corresponding reported device performance metrics in the format of a table. Instead, it relies on verification and validation (V&V) testing against system requirements and a comparison to a predicate device to demonstrate safety and effectiveness.

The "Nonclinical Testing" section (Section 7) describes the general approach to validating the device:

• Acceptance Criteria (Implied): The software device must meet its system requirements. Specifically for the image noise reduction feature, the algorithm must decrease image noise without adding artifacts and produce reproducible filtering results.
• Reported Device Performance:
  • V&V test procedures were created and executed on the software using an NTSC analog video input source.
  • The results were compiled into V&V test reports, which presumably showed that the system requirements were met.
  • Planned risk mitigations in the hazard analysis were verified.
  • The image noise reduction feature was verified to ensure it produced safe and effective results, specifically that image noise was decreased without adding image artifacts and that filtering results were reproducible.

Since no specific numerical acceptance criteria or performance metrics are provided, a table of acceptance criteria and reported device performance cannot be generated. The submission emphasizes that all system requirements were met and verified through testing.

Study Details

1. Sample size used for the test set and the data provenance:

   • The submission mentions "exam-specific imagery" and "colon images" were used for testing the image noise reduction feature, and an "NTSC analog video input source" for general V&V. However, the exact sample size (number of images/videos/exams) for the test set is not specified.
   • Data provenance is not explicitly stated, but the use of an "NTSC analog video input source" suggests a laboratory or controlled setting for general testing. For "colon images," it's unclear if these were from a specific country or whether they were retrospective or prospective.

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

   • The submission does not mention the involvement of experts to establish ground truth for the test set. The validation appears to be primarily engineering-based, comparing the software's output to defined functional requirements for video capture, archiving, noise reduction, etc.

3. Adjudication method for the test set:

   • No adjudication method is described as the ground truth was not established by multiple experts.

4. 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. The device's purpose is as an accessory for viewing, recording, and managing colonoscopy data, not primarily for diagnostic interpretation or aiding human readers in decision-making in a way that would require an MRMC study. Its function is to streamline the workflow and manage visual data.

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

   • While the device is a "software application" and performs functions like digital noise reduction automatically, the submission does not present data specifically illustrating "standalone" diagnostic performance in the way an AI-powered diagnostic tool would. Its functions are assistive to a human-performed procedure (colonoscopy) rather than providing independent diagnostic conclusions. The noise reduction is an algorithmic standalone function, but its "performance" is verified against quality improvement criteria (decreased noise, no artifacts) rather than diagnostic accuracy.

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

   • The description implies a functional ground truth based on system requirements and expected output. For example, for noise reduction, the ground truth is "decreased image noise without adding image artifacts" and "reproducible filtering results," which would be assessed by visual inspection and technical evaluation rather than, for instance, pathology reports for diagnostic accuracy.

7. The sample size for the training set:

   • This submission describes a software accessory for data management and image processing, not a machine learning model that requires a "training set" in the conventional sense of supervised or unsupervised learning. Therefore, no training set sample size is applicable or provided.

8. How the ground truth for the training set was established:

   • As there is no mention of a training set, the establishment of its ground truth is also not applicable or provided.

Ask a specific question about this device