Search Results

AISight Dx is a software only device intended for viewing and management of digital images of scanned surgical pathology slides prepared from formalin-fixed paraffin embedded (FFPE) tissue. It is an aid to the pathologist to review, interpret, and manage digital images of these slides for primary diagnosis. AISight Dx is not intended for use with frozen sections, cytology, or non-FFPE hematopathology specimens.

It is the responsibility of a qualified pathologist to employ appropriate procedures and safeguards to assure the quality of the images obtained and, where necessary, use conventional light microscopy review when making a diagnostic decision. AISight DX is intended to be used with interoperable displays, scanners and file formats, and web browsers that have been 510(k) cleared for use with the AISight Dx or 510(k)-cleared displays, 510(k)-cleared scanners and file formats, and web browsers that have been assessed in accordance with the Predetermined Change Control Plan (PCCP) for qualifying interoperable devices.

AISight Dx is a web-based, software-only device that is intended to aid pathology professionals in viewing, interpretation, and management of digital whole slide images (WSI) of scanned surgical pathology slides prepared from formalin-fixed, paraffin-embedded (FFPE) tissue obtained from Hamamatsu NanoZoomer S360MD Slide scanner or Leica Aperio GT 450 DX scanner (Table 1). It aids the pathologist in the review, interpretation, and management of pathology slide digital images used to generate a primary diagnosis.

Here's a breakdown of the acceptance criteria and the study details for the AISight Dx device, based on the provided FDA 510(k) Clearance Letter:

Acceptance Criteria and Reported Device Performance

Study Details

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

   • The document states that two separate clinical studies were conducted, one for each scanner (Hamamatsu NanoZoomer S360MD and Leica Aperio GT 450 DX).
   • The sample sizes for these clinical studies are not explicitly stated in the provided text.
   • Data Provenance: Not explicitly mentioned, but the study compares performance against "the original sign-out pathologic diagnosis using MO [ground truth, (GT)] rendered at the institution," suggesting the data is derived from clinical practice, likely retrospective or a mix, given the "original sign-out" aspect. The country of origin is not specified.

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

   • The study involved "3 reading pathologists" for assessing the differences in major discordance rates.
   • Qualifications of experts: Not explicitly stated, but they are referred to as "reading pathologists," indicating they are qualified to make primary diagnoses.

3. Adjudication method for the test set:

   • The "reference (main) diagnosis" was the "original sign-out pathologic diagnosis using MO [ground truth, (GT)] rendered at the institution."
   • The document implies that this "original sign-out" acted as the ground truth. There's no explicit mention of an adjudication process (e.g., 2+1, 3+1 consensus) to establish this ground truth beyond the initial clinical diagnosis. The major discordance rate was calculated between MD (manual digital read), MO (manual optical read), and GT (original sign-out).

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:

   • Yes, an MRMC-like study was done as it involved "3 reading pathologists" evaluating cases using both manual digital (MD) and manual optical (MO) methods.
   • Effect size of improvement with AI vs without AI assistance: This study did not measure the improvement of human readers with AI assistance. The AISight Dx is presented as a viewer and management software, not an AI-assisted diagnostic tool. The study aimed to demonstrate non-inferiority of digital viewing (MD) versus traditional optical viewing (MO) for primary diagnosis, where the software is simply the viewing platform, not an aid in interpretation itself. Therefore, no "effect size of human readers improving with AI vs without AI assistance" is reported because the device is not described as providing AI assistance for diagnostic tasks in this context.

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

   • No. The AISight Dx is explicitly described as "an aid to the pathologist to review, interpret, and manage digital images." The clinical study evaluated "manual digital read (MD)" which is a human pathologist reading digital slides using the AISight Dx, compared to "manual optical (MO)" which is a human pathologist reading glass slides. The device is not an autonomous AI diagnostic algorithm.

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

   • The ground truth (GT) for the clinical study was the original sign-out pathologic diagnosis using manual optical microscopy (MO) rendered at the institution. This can be categorized as a form of expert (pathologist) ground truth based on clinical practice/standard of care.

7. The sample size for the training set:

   • The document for AISight Dx does not mention a training set size. This is expected as AISight Dx is described as a viewing and management software, not an AI model that requires a training set for diagnostic capabilities. The performance data focuses on its function as a display and interpretation platform for human pathologists.

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

   • As there's no mention of a training set for an AI model within the AISight Dx software (it's a viewer), there's no information on how a ground truth for such a set would be established.

Ask a specific question about this device

Novo is a software only device intended for viewing and management of digital images of scanned surgical pathology slides prepared from formalin-fixed paraffin embedded (FFPE) tissue. It is an aid to the pathologist to review, interpret, and manage digital images of these slides for primary diagnosis. Novo is not intended for use with frozen sections, cytology, or non- FFPE hematopathology specimens.

It is the responsibility of a qualified pathologist to employ appropriate procedures and safeguards to assure the quality of the images obtained and, where necessary, use conventional light microscopy review when making a diagnostic decision. Novo is intended for use with the Philips Ultra Fast Scanner and the Barco PP27QHD or Philips PS27QHDCR display.

The PathAI Novo device is a web-based software-only device that is intended to aid pathology professionals in the viewing, interpretation, and management of digital whole slide images (WSIs) of scanned surgical pathology slides prepared from formalin-fixed paraffin embedded (FFPE) tissue using the Philips IntelliSite Pathology Solution (PIPS) Ultra Fast Scanner (UFS).

The proposed device is typically operated as follows:

• 1. A user prepares and scans slides and reviews the slide quality in accordance with the PIPS UFS IFU and standard lab procedures. The Novo device workflow is initiated when a user uploads WSIs from the local file system to the cloud storage using Novo.
• 2. After uploading WSIs to cloud storage using Novo, a user builds a patient accession using the patient's medical record number (MRN), date of birth (DOB) and accession ID to support linkage of one or more slides from a single procedure using patient identifiers in Novo.
• 3. A pathologist uses the slide viewer to perform their primary diagnosis workflow including zooming and panning images.

After viewing all images belonging to a particular accession, the pathologist will make a diagnosis.

The provided text describes the regulatory clearance for the "Novo" device, a software-only whole slide imaging system, and references a clinical study conducted to establish its substantial equivalence to a predicate device. However, the document primarily focuses on regulatory approval and does not contain the detailed acceptance criteria table or comprehensive study breakdown as requested in the prompt.

Therefore, the following response will extract what is available and highlight where information is missing based on your request.

Acceptance Criteria and Device Performance for Novo (as described by available information)

Based on the provided FDA 510(k) summary, details regarding specific quantifiable acceptance criteria and performance beyond a non-inferiority finding are limited. The document focuses on demonstrating substantial equivalence to a predicate device (Philips IntelliSite Pathology Solution - PIPS).

Table of Acceptance Criteria and Reported Device Performance

Missing Information (Not Available in the Provided Text):

• Explicit, pre-specified quantitative acceptance criteria beyond the non-inferiority margin for major discordance.
• Detailed quantitative performance for features like color differences beyond a statement of similarity to JPEG-compressed images.

Study Details (Extracted from the provided text):

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

• Test Set Sample Size: Not explicitly stated. The text mentions "WSIs of H&E stained FFPE tissue slides."
• Data Provenance: Not specified for the test set. (e.g., country of origin, retrospective/prospective).

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

• Number of Experts: Not specified.
• Qualifications of Experts: Not specified. The reference diagnosis (ground truth) was the "original sign-out pathologic diagnosis using MO [manual optical] rendered at the institution," implying involvement of qualified pathologists for routine diagnosis, but specific qualifications for ground truth establishment are not given.

4. Adjudication method for the test set:

• Adjudication Method: Not specified. The study compared "major discordance rates between MD [manual digital read, using Novo] and MO [manual optical] when compared to the reference (main) diagnosis, which was the original sign-out pathologic diagnosis using MO [ground truth, (GT)] rendered at the institution." This suggests a comparison against an existing diagnosis, not necessarily a separate adjudication process for the test set.

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:

• MRMC Study: Yes, a clinical study was conducted. It involved human pathologists making diagnoses using both the Novo device (MD) and conventional microscopy (MO).
• Effect Size of Improvement: The study focused on non-inferiority of the manual digital read (MD) using Novo compared to manual optical (MO) read. It did not measure "how much human readers improve with AI vs without AI assistance" because Novo is described as a "viewing and management" tool, not an AI-assisted diagnostic tool. Its purpose is to present the image for the pathologist's primary diagnosis, making it a replacement for conventional microscopy, not an enhancement tool for human readers in the context of an AI-assisted workflow. The primary outcome was maintaining diagnostic accuracy (non-inferiority) when switching from MO to MD.

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

• Standalone Performance: No, this device is a "viewer and management" software for human pathologists. It does not perform diagnostic algorithms independently. The performance data presented (non-inferiority) is explicitly human-in-the-loop (MD: "manual digital read").

7. The type of ground truth used:

• Ground Truth Type: "The original sign-out pathologic diagnosis using MO [manual optical] rendered at the institution." This can be categorized as expert consensus/clinical outcomes data based on standard practice.

8. The sample size for the training set:

• Training Set Sample Size: Not applicable/Not specified. Novo is a viewing and management system, not described as an AI/machine learning algorithm that requires a "training set" in the conventional sense for a diagnostic prediction model. The "pixel-wise comparison" tests were likely technical assessments, not dependent on a "training set."

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

• Ground Truth Establishment for Training Set: Not applicable, as there is no mention of an algorithm requiring a "training set" with ground truth in the provided information.

Ask a specific question about this device