Search Results

The software is to be used for the remote viewing of files generated by a medical scanning device and acquired according to the dominant industry-standard communications format (DICOM 3.0).

The software is intended to provide the means for medical professionals to display data generated by medical scanning devices on a personal computer or workstation.

AMICAS is an integrated client-server software system designed to allow access to medical inages by radiologists, referring physicians and other licensed professionals. This product is intended to allow the review of images on a digital Picture Archive and Communication System (PACS) network using a personal computer or workstation configured for standard Internet access. The server component of AMICAS is installed on a computer configured with connections to both the PACS and the Internet. Typically, images will be accessed through a World Wide Web (WWW) browser (the client) such as Microsoft Internet Explorer or Netscape Navigator. Radiology workstations can Query and Retrieve images from AMICAS using standard DICOM protocols.

AMICAS can be used for image distribution within a hospital, a managed care organization or an isolated imaging center. It can also serve as a telemedicine link between widely separate organizations.

Here's a breakdown of the acceptance criteria and study information for the AMICAS Web/Intranet Image Server based on the provided document:

This device is not an AI/ML powered device, so several of the requested sections (e.g., MRMC study, standalone performance) are not applicable. The device is an image communication device, essentially a PACS viewer, that utilizes lossy compression. The primary "study" described is a clinical collaboration and experience, rather than a formal pre-market clinical trial as one might expect for a diagnostic AI device today.

Acceptance Criteria and Reported Device Performance

Given that this is a 510(k) submission from 1997 for an image communication device (PACS viewer) that utilizes lossy compression, the concept of "acceptance criteria" is primarily focused on demonstrating equivalence to predicate devices and ensuring the compressed images retain sufficient diagnostic quality. The document doesn't present explicit "acceptance criteria" in a typical numerical format for diagnostic accuracy or sensitivity/specificity. Instead, it relies on demonstrating that its image compression ratios are "consistent with" and "substantially equivalent to" those used in already cleared devices and clinical practice.

Study Details

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

   • Test Set Sample Size: Not explicitly stated as a separate "test set" for this specific device's evaluation. The submission refers to a "three years of clinical experience" at Massachusetts General Hospital (MGH) with wavelet compression, which served as the basis for setting target compression ratios.
   • Data Provenance: Massachusetts General Hospital (MGH), Boston, MA. This would be considered retrospective experience with the underlying compression technology, not a prospective trial specifically for AMICAS itself. The cited academic reference (Goldberg et al. 1993) involved a prospective study of 685 transmitted clinical cases but was conducted on a different teleradiology system using wavelet compression, serving as an academic reference for the technology itself, not direct evidence for AMICAS.

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

   • For the "three years of clinical experience" at MGH, the "experts" were the clinical radiologists and practitioners at the hospital who used the wavelet compression technology in their daily workflow. No specific number is provided, nor strict qualifications beyond being "radiologists" as implied by the context of "Massachusetts General Hospital (MGH), Boston, MA" and "collaborated with the Radiology Associates of Massachusetts General Hospital (MGH)".
   • For the Goldberg et al. (1993) reference, the study involved radiologists reading transmitted images, but specific numbers and qualifications related to ground truth establishment for their study are not detailed in this document.

3. Adjudication method for the test set:

   • Not applicable/Not described. The "study" here is more of a validation of the compression ratios based on real-world clinical use and equivalence to a predicate device, rather than a formal diagnostic accuracy study with a specific adjudication process.

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. This is a 1997 submission for a PACS viewer (an image communication device), not an AI/ML powered device. Therefore, no MRMC study comparing human readers with and without AI assistance was conducted.

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

   • Yes, in a sense, as the algorithm's performance (image compression ratios and visual quality post-compression) was evaluated "alone" in the context of its output, but not as a diagnostic algorithm. The "test data and conclusions" primarily refer to the automatic selection of target compression ratios by AMICAS. The document explicitly states: "The software... It does not provide a diagnosis. It provides information/data only."

6. The type of ground truth used:

   • For the "three years of clinical experience" at MGH, the ground truth was implicitly clinical diagnosis and radiologist interpretation of images compressed using wavelet technology, established over time in routine clinical practice. The absence of reported issues with diagnostic accuracy at those compression ratios informed the "acceptance."
   • The Goldberg et al. 1993 reference abstract suggests diagnostic accuracy of transmitted clinical cases.

7. The sample size for the training set:

   • Not applicable/Not explicitly defined. This device does not use a "training set" in the machine learning sense. The "training" for the compression ratios was derived from "three years of clinical experience with wavelet compression at Massachusetts General Hospital," which implies a very large, ongoing dataset of real-world clinical images over that period.

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

   • Not applicable/Not explicitly defined. As there is no "training set" in the ML sense, there's no ground truth established for it. The general "ground truth" for the clinical experience that informed the compression ratios was the standard of care for radiological diagnosis at MGH.

Ask a specific question about this device