(109 days)
The syngo.CT Lung CAD VC30 device is a Computer-Aided Detection (CAD) tool designed to assist radiologists in the detection of solid pulmonary nodules during review of multi-detector computed tomography (MDCT) examinations of the chest. The software is an adjunctive tool to alert the radiologist to regions of interest that may have been initially overlooked. The syngo.CT Lung CAD device is intended to be used as a second reader after the radiologist has completed his/her initial read.
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 ≥ 3.0 mm in size. The device processes images acquired with Siemens multi-detector CT scanners with 4 or more detector rows.
The syngo.CT Lung CAD device supports the full range of nodule locations (central, peripheral) and contours (round, irregular). The detection performance of the syngo.CT Lung CAD device is optimized for nodules between 3.0 mm and 20.0 mm in size.
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 for use in a second reader mode. syngo MM Oncology (FDA clearance K191309) is implemented on the syngo.via platform (FDA clearance K191040), 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 is intended to be used as a second reader only after the initial read is completed.
The subject device and the predicate device has the same basic technical characteristics as the predicate; however, the fundamental technology has been replaced by deep learning technology. Specifically, the predicate VC20 uses feature-based and Machine Learning whereas the current VC30 uses algorithms based on Convolutional Neural Networks. This does not introduce new types of safety or effectiveness concerns. In particular, as demonstrated by the statistical analysis and results of the standalone benchmark evaluations:
i. The standalone accuracy has been shown not only to be non-inferior but actually superior to that of the device and
ii. The marks generated by the two devices have been shown to be reasonably consistent.
This device description holds true for the subject device, syngo.CT Lung CAD, software version VC30, as well as the predicate device, syngo.CT Lung CAD, software version VC20.
Here's a breakdown of the acceptance criteria and the study proving the device meets them, based on the provided FDA 510(k) summary for syngo.CT Lung CAD (VC30):
Device Name: syngo.CT Lung CAD (VC30)
Intended Use: A Computer-Aided Detection (CAD) tool to assist radiologists in the detection of solid pulmonary nodules (≥ 3.0 mm) during review of multi-detector computed tomography (MDCT) examinations of the chest. It's an adjunctive tool to alert radiologists to initially overlooked regions of interest, used as a second reader after the radiologist's initial read.
1. Table of Acceptance Criteria and Reported Device Performance
The document primarily focuses on demonstrating non-inferiority and superiority to the predicate device rather than explicitly stating acceptance criteria with numerical targets for metrics like sensitivity or specificity. However, based on the conclusions regarding "standalone accuracy" and "false positive rate," we can infer the implicit criteria and the reported performance as comparative to the predicate.
| Acceptance Criteria (Inferred from comparison to predicate) | Reported Device Performance (syngo.CT Lung CAD VC30) |
|---|---|
| Standalone accuracy (sensitivity for nodule detection) is non-inferior to predicate (syngo.CT Lung CAD VC20). | Superior to predicate (syngo.CT Lung CAD VC20). |
| False positive rate is not worse than predicate (syngo.CT Lung CAD VC20). | Improved (reduced) compared to predicate (syngo.CT Lung CAD VC20). |
| Consistency of marks (location and extent) with predicate (syngo.CT Lung CAD VC20). | Reasonably consistent with marks produced by predicate (syngo.CT Lung CAD VC20). |
Note: The document describes the study as a "standalone benchmark evaluation" focused on comparing VC30's performance to VC20's. Specific numerical metrics for sensitivity, specificity, or FPs are not provided in this summary, but the conclusions about superiority and reduction in FPs serve as the performance statement.
2. Sample Size Used for the Test Set and Data Provenance
- Sample Size for Test Set: The document states that "The endpoints to establish meaningful and statistically relevant performance and equivalence of the device and sample size were considered and defined as part of the test protocols." However, the specific number of cases or nodules in the test set is not provided in this summary.
- Data Provenance: Not explicitly stated regarding country of origin. The document mentions "Non-clinical performance testing was performed at various levels for verification and validation of the device intended use and to ensure safety and effectiveness." It does not specify if the data was retrospective or prospective.
3. Number of Experts Used to Establish Ground Truth and Qualifications
- Number of Experts: Not specified in the provided text.
- Qualifications of Experts: Not specified in the provided text.
4. Adjudication Method for the Test Set
- Adjudication Method: Not specified in the provided text. The document refers to "ground truth" but does not detail the method by which it was established.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
- MRMC Study: The document does not mention a multi-reader multi-case (MRMC) comparative effectiveness study comparing human readers with AI vs. without AI assistance. The study described is a "standalone benchmark evaluation" comparing the performance of the new AI algorithm (VC30) to the previous algorithm (VC20).
- Effect Size of Human Improvement: Not applicable, as no MRMC study is detailed here.
6. Standalone (Algorithm Only) Performance Study
- Standalone Study: Yes, a standalone study was done. The document explicitly states: "The standalone performance test proved that the standalone sensitivity of syngo.CT Lung CAD VC30 is superior to that of syngo.CT Lung CAD VC20 (predicate) and the false positive rate improved (reduced)."
7. Type of Ground Truth Used
- Type of Ground Truth: The document refers to "ground truth" for the test set, stating that it was established to define "meaningful and statistically relevant performance." However, the specific method (e.g., expert consensus, pathology, follow-up outcomes) for establishing this ground truth is not detailed in the provided summary.
8. Sample Size for the Training Set
- Sample Size for Training Set: The document does not provide the sample size used for the training set. It only mentions that the "fundamental technology has been replaced by deep learning technology," indicating a training process was involved.
9. How the Ground Truth for the Training Set Was Established
- How Ground Truth for Training Set Was Established: The document does not provide details on how the ground truth for the training set was established. It only describes the functional components of the new syngo.CT Lung CAD VC30 as using Convolutional Neural Networks (CNN) for lung segmentation, candidate generation, feature calculation, and candidate classification, which inherently require labeled training data.
§ 892.2050 Medical image management and processing system.
(a)
Identification. A medical image management and processing system is a device that provides one or more capabilities relating to the review and digital processing of medical images for the purposes of interpretation by a trained practitioner of disease detection, diagnosis, or patient management. The software components may provide advanced or complex image processing functions for image manipulation, enhancement, or quantification that are intended for use in the interpretation and analysis of medical images. Advanced image manipulation functions may include image segmentation, multimodality image registration, or 3D visualization. Complex quantitative functions may include semi-automated measurements or time-series measurements.(b)
Classification. Class II (special controls; voluntary standards—Digital Imaging and Communications in Medicine (DICOM) Std., Joint Photographic Experts Group (JPEG) Std., Society of Motion Picture and Television Engineers (SMPTE) Test Pattern).