Search Results

syngo.CT Lung CAD device is a computer-aided detection (CAD) tool designed to assist radiologists in the detection of solid and subsolid pulmonary nodules during review of multi-detector computed tomography (MDCT) from multivendor examinations of the chest. The software is an adjunctive tool to alert the radiologist to regions of interest (ROI) that may be otherwise overlooked.

The syngo.CT Lung CAD device may be used as a concurrent first reader followed by a full review of the case by the radiologist or as second reader after the radiologist has completed his/her initial read.

The syngo.CT Lung CAD device may also be used in "solid-only" mode, where potential (or suspected) sub-solid and/or fully calcified CAD findings are filtered out.

The software device is an algorithm which does not have its own user interface component for displaying of CAD marks. The Hosting Application incorporating syngo. CT Lung CAD is responsible for implementing a user interface.

Siemens Healthcare GmbH intends to market the syngo.CT Lung CAD which is a medical device that is designed to perform CAD processing in thoracic CT examinations for the detection of solid pulmonary nodules (between 3.0 mm and 30.0mm) and subsolid nodules (between 5.0 mm and 30.0mm) in average diameter. The device processes images acquired with multi-detector CT scanners with 16 or more detector rows recommended.

The syngo.CT Lung CAD device supports the full range of nodule locations (central, peripheral) and contours (round, irregular).

The syngo.CT Lung CAD sends a list of nodule candidate locations to a visualization application, such as syngo MM Oncology, or a visualization rendering component, which generates output images series with the CAD marks superimposed on the input thoracic CT images to enable the radiologist's review. syngo MM Oncology (FDA clearanceK211459 and subsequent versions ) is deployed on the syngo.via platform (FDA clearance K191040 and subsequent versions), which provides a common framework for various other applications implementing specific clinical workflows (but are not part of this clearance) to display the CAD marks. The syngo.CT Lung CAD device may be used either as a concurrent first reader, followed by a review of the case, or as a second reader only after the initial read is completed

The provided text describes the Siemens syngo.CT Lung CAD (Version VD30) and its substantial equivalence to its predicate device (syngo.CT Lung CAD Version VD20). The primary change in VD30 is the introduction of a "solid-only" mode. The acceptance criteria and study details are primarily focused on demonstrating that the VD30 in "solid-only" mode is not inferior to VD20 in standard mode, and that VD30 in standard mode is not inferior to VD20 in standard mode. Since the document primarily focuses on demonstrating non-inferiority to a predicate device, explicit acceptance criteria values (e.g., minimum sensitivity thresholds) are not explicitly stated as numerical targets. Instead, the acceptance is based on statistical non-inferiority.

Here's a breakdown of the requested information based on the provided text:

1. A table of acceptance criteria and the reported device performance

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

• Sample Size: 712 CT thoracic cases.
• Data Provenance: Retrospectively collected data from 3 sources:
  • The UCLA study (232 cases)
  • The original PMA study (145 cases)
  • Additional cases (335 cases)

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

The document differentiates ground truth establishment based on the data source:

• UCLA data: Reference standard (ground truth) was determined as part of the reader study for the predicate device (K203258). The number and qualifications of experts are not explicitly stated for this subset in the provided text for VD30, but it refers to the predicate clearance.
• PMA study cases: 18 readers were used. Qualifications are not explicitly stated, but 9 of the 18 readers were needed for declaring a true nodule.
• Additional cases: 7 readers were used. Qualifications are not explicitly stated, but 4 of the 7 readers were needed for declaring a true nodule.

4. Adjudication method (e.g., 2+1, 3+1, none) for the test set

The adjudication method varied based on the data source:

• PMA study cases: 9 out of 18 readers were needed for declaring a true nodule. This suggests a majority consensus from a large panel.
• Additional cases: 4 out of 7 readers were needed for declaring a true nodule. This also suggests a majority consensus.
• UCLA data: "Reference standard for the UCLA data was determined as part of the reader study (K203258)." Specific adjudication details for this subset are not provided in this document but are referenced to the predicate device's clearance.

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

A MRMC comparative effectiveness study involving human readers with and without AI assistance is not explicitly described in this document. The statistical analysis performed was a standalone performance analysis to demonstrate substantial equivalence between two CAD versions (VD30 vs VD20), focusing on the algorithm's performance metrics (sensitivity, FPs).

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

Yes, a standalone performance analysis was done. The document states: "The standalone performance analysis was designed to demonstrate the substantial equivalence between syngo.CT Lung CAD VD30A (VD30) and the predicate device syngo.CT Lung CAD VD20."

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

The ground truth was established through expert consensus/reader review.

• For PMA cases: 9 out of 18 readers' consensus.
• For additional cases: 4 out of 7 readers' consensus.
• For UCLA data: Reference standard from a reader study.

8. The sample size for the training set

The document does not explicitly state the sample size for the training set. It mentions that the algorithm is based on Convolutional Networks (CNN) and that the lung segmentation algorithm for VD30 in particular is "trained on lung CT data including comorbidities for robustness," but the specific number of cases for this training set is not provided.

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

The document does not explicitly describe how the ground truth for the training set was established. It only states that the lung segmentation algorithm was "trained on lung CT data" and that the overall algorithm uses CNNs, implying supervised learning, which would require ground truth annotations. However, the method of obtaining these annotations is not detailed.

Ask a specific question about this device