Search Results

The Discovery XR656 HD is intended to generate digital radiographic images of the skull, spinal column, chest, abdomen. extremities, and other body parts in patients of all ages. Applications can be performed with the patient sitting, standing. or lying in the prone or supine position and the system is intended for use in all routine radiography exams. Optional image pasting function enables the operator to stitch sequentially acquired radiographs into a single image.

The Discovery XR656 HD incorporates AutoGrid, which is an optional image processing software installed as a part of the systems Helix image processing software. AutoGrid can be used in lieu of an anti-scatter grid to improve image contrast in general radiographic images by reducing the effects of scatter radiation.

When the VolumeRAD option is included on the system can generate tomographic images of human anatomy including the skull, spinal column, chest, abdomen, extremities, and other body parts in patients of all ages.

When the VolumeRAD option is used for patients undergoing thoracic imaging, it is indicated for the detection of lung nodules. VolumeRad generates diagnostic images of the radiologist in achieving superior detectability of lung nodules versus posterior and left lateral views of the chest, at a comparable radiation level.

The device is not intended for mammographic applications.

The Discovery XR656 HD Radiography X-ray System is designed as a modular system with components that include an Overhead Tube Suspension with tube/collimator, wallstand, Table, X-ray generator, and cleared wireless digital detectors. The list of detectors verified and validated for use with the Discovery XR656 HD system, including their specifications, are provided in the user documentation. The System generates diagnostic radiographic images which can be sent through a DICOM network for applications including printing, viewing, and storage.
The components may be combined in different configurations to meet specific customer needs. In addition, upgrade configurations are available for predicate devices.
The optional image pasting function enables the operator to stitch sequentially acquired radiographs into a single image.
This 510(k) is to incorporate the VolumeRad advanced application that was currently available on the Discovery XR656 product onto the Discovery XR656 HD, as well as introduce a new Metal Artifact Reduction Algorithm, and an optional standalone console to take any Helix™ acquired images via DICOM (such as from a Discovery XR656 HD, Optima XR646 HD, or Optima XR240amx) and process the images independently of the system it was acquired on.

This document is a 510(k) Premarket Notification submission for the GE Healthcare Discovery XR656 HD with VolumeRad. The submission details the device's technical characteristics, intended use, and a comparison to predicate and reference devices to establish substantial equivalence.

Based on the provided text, the device itself (Discovery XR656 HD with VolumeRad) is an X-ray system, not an AI or algorithm. Therefore, the questions related to AI performance metrics such as reader improvement with AI assistance, standalone algorithm performance, and sample sizes for training/test sets specifically for an AI component are not directly applicable.

However, the document does describe the "Metal Artifact Reduction algorithm for VolumeRad" and mentions its evaluation. This suggests an algorithmic component, though not an AI in the common sense of machine learning for diagnosis. The data provided focuses on demonstrating substantial equivalence to predicate devices for the overall system and its features, including the VolumeRad function with updated detectors and the metal artifact reduction algorithm.

Here's an analysis based on the information available, addressing the relevant points:

1. Table of Acceptance Criteria and Reported Device Performance:

The document primarily focuses on establishing substantial equivalence for the Discovery XR656 HD with VolumeRad to predicate devices, rather than defining specific acceptance criteria for a new AI algorithm and reporting its performance against those. The "performance" discussed is related to the overall system's safety and effectiveness, and the ability of the VolumeRad feature to generate diagnostic images comparable to or better than traditional views for lung nodule detection.

The statement regarding VolumeRad: "VolumeRad generates diagnostic images of the radiologist in achieving superior detectability of lung nodules versus posterior and left lateral views of the chest, at a comparable radiation level." acts as a performance claim for the VolumeRad feature itself, which is part of the device.

2. Sample Size Used for the Test Set and Data Provenance:

• Test Set for VolumeRad feature and Metal Artifact Reduction Algorithm: The document states that "bench testing using anthropomorphic phantoms was sufficient" for evaluating the Metal Artifact Reduction algorithm and for showing the equivalence of the VolumeRad feature with updated resolution detectors.
  • Sample Size: Not explicitly stated as a number of cases or patients from a clinical study for the test set. It refers to "anthropomorphic phantoms."
  • Data Provenance: Not human clinical data. The data originates from "anthropomorphic phantoms" used in bench testing. Given it's a GE Healthcare product, typically such testing occurs internally or at partner facilities. The location of the manufacturer is China.
  • Retrospective/Prospective: Not applicable as it's bench testing with phantoms.

3. Number of Experts Used to Establish Ground Truth for the Test Set and Qualifications:

• Since the testing was primarily bench testing with anthropomorphic phantoms, there is no mention of human experts establishing ground truth in the context of reading images from a test set. Evaluation would likely involve technical measurements and visual assessment by product development engineers or possibly consulting radiologists for image quality, but this isn't described as a formal ground truth process for a clinical test set.

4. Adjudication Method for the Test Set:

• Not applicable, as the evaluation was primarily bench testing with phantoms.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study:

• No, a multi-reader multi-case (MRMC) comparative effectiveness study comparing human readers with and without AI assistance was not done. The submission explicitly states: "The subject of this premarket submission, Discovery XR656 HD with VolumeRad, did not require clinical studies to support substantial equivalence for the changes identified."
• Effect Size: Not determined, as no such study was performed.

6. Standalone (Algorithm Only Without Human-in-the-Loop Performance) Study:

• The document implies that the fundamental algorithm to create the VolumeRad image set is identical to the algorithm cleared under K132261. The Metal Artifact Reduction algorithm was evaluated via bench testing. While these are algorithmic components, the overall "device" is an X-ray system. The performance claims for VolumeRad are implicitly related to its ability to present images that aid the radiologist (human-in-the-loop). Bench testing of the algorithms was done, but not as a standalone diagnostic AI performance study in the typical sense for clinical claims.

7. Type of Ground Truth Used:

• For the technical evaluation of the VolumeRad feature and the Metal Artifact Reduction algorithm, the "ground truth" was established through bench testing using anthropomorphic phantoms. This means known conditions (e.g., presence/absence of nodules, specific metal artifacts) were simulated in the phantoms to assess the system's output.

8. Sample Size for the Training Set:

• The document does not describe the development of a new AI algorithm that would typically involve a "training set." The VolumeRad algorithm is stated to be "identical" to a previously cleared algorithm. The Metal Artifact Reduction algorithm is new, but its development process (including any training data if it were a machine learning algorithm) is not detailed. Therefore, the sample size for a training set is not provided.

9. How the Ground Truth for the Training Set Was Established:

• Not applicable as no specific "training set" for a new AI algorithm is described. For the general development of the overall system and its included algorithms, ground truth would be established through engineering specifications, phantom studies for image quality, and comparison against known physical properties.

Ask a specific question about this device