Search Results
Found 1 results
510(k) Data Aggregation
(128 days)
Constellation is intended for non-invasive labeling and calculation of quantitative measurements for anatomical regions. Constellation utilizes DICOM MR images gathered on a GE MR450W that encompass the whole-body. It is intended to be used for healthy adult patients. Clinicians may use Constellation as a clinical decision support tool, but it is not to be used in triage events, emergency medicine, or critical care. A clinician retains the ultimate responsibility for making the pertinent diagnosis based on their standard practices.
Constellation is an automated image post-processing software application used in a clinical MRI setting. Constellation combines MR images from overlapping anatomical stations to segment anatomical regions and provide associated quantitative measurements for whole-body patient anatomy. These anatomical regions include the lower limb muscles, visceral adipose tissue, subcutaneous adipose tissue, kidneys, liver, spleen, lungs, and brain structures. A PDF report contains quantified measurements alonqside a whole-body visualization and segmented label images.
Constellation provides alpha-blending of the anatomical image with the corresponding labels in the final report. This process combines one stitched output (background) with another (foreground) to create a final anatomical label with both grayscale and color resulting from the blending of the background (greyscale) and foreground label.
Constellation is a tool intended to assist trained physicians in the assessment of MR images of the whole body. The software is not designed to provide any automated detection or diagnosis. Physicians retain the ultimate responsibility for making any diagnosis from the presented images based on their standard practices and patient background, clinical history, symptoms, and other diagnostic information.
Here's a breakdown of the acceptance criteria and the study details for the Constellation (CON-001) device, based on the provided text:
Acceptance Criteria and Device Performance Study
The Constellation device, an automated image post-processing software for whole-body MR images, underwent various performance tests to demonstrate its safety and effectiveness. The primary metrics used for evaluating segmentation accuracy were the Dice Similarity Coefficient (DSC) and mean percent absolute difference.
1. Table of Acceptance Criteria and Reported Device Performance
The document does not explicitly state acceptance criteria with specific numerical thresholds for each metric. Instead, it describes what was evaluated and the metrics used. The conclusion states that the device "performs to its intended use" and "is as safe, as effective, and performs as well as the predicate devices." Without specific numerical targets or results mentioned in the provided text, the "Reported Device Performance" column reflects what was measured rather than specific values achieved against a predefined threshold.
| Metric/Evaluation Area | Acceptance Criteria (as implied/evaluated) | Reported Device Performance (as stated) |
|---|---|---|
| Segmentation Accuracy (Lungs, Liver, Spleen, Kidneys, Muscle, Visceral & Subcutaneous Fat) | Performance deemed acceptable for clinical use, comparable to predicate devices. | Evaluated using Dice Similarity Coefficient (DSC) as the primary figure of merit (FOM) and mean percent absolute difference as the secondary FOM. The test results demonstrated that Constellation performs to its intended use and does not introduce new questions of safety or effectiveness. The device is claimed to be "as safe, as effective, and performs as well as the predicate devices." |
| Segmentation Accuracy (Brain Cortical/Subcortical Regions) | Performance deemed acceptable for clinical use, comparable to predicate devices. | Evaluated via mean percent absolute difference as the primary FOM and Pearson's correlation coefficient as the secondary FOM. The test results demonstrated that Constellation performs to its intended use and does not introduce new questions of safety or effectiveness. The device is claimed to be "as safe, as effective, and performs as well as the predicate devices." |
| Liver Volume of Interest (VOI) Placement | Placement accuracy acceptable per expert adjudication. | Evaluated via majority voting of three radiologists. The test results demonstrated that Constellation performs to its intended use and does not introduce new questions of safety or effectiveness. The device is claimed to be "as safe, as effective, and performs as well as the predicate devices." |
| Device Repeatability (same MRI scans passed twice) | Consistent results; acceptable DSC and mean percent absolute difference. | Evaluated by calculating the DSC and mean percent absolute difference for the same MRI scans passed twice through Constellation. The test results demonstrated that Constellation performs to its intended use and does not introduce new questions of safety or effectiveness. The device is claimed to be "as safe, as effective, and performs as well as the predicate devices." |
| Test-Retest Measurement Repeatability | Consistent measurements; acceptable mean percent absolute difference and Pearson's correlation coefficient. | Evaluated using the mean percent absolute difference and Pearson's correlation coefficient as the primary and secondary FOMs. The test results demonstrated that Constellation performs to its intended use and does not introduce new questions of safety or effectiveness. The device is claimed to be "as safe, as effective, and performs as well as the predicate devices." |
| Manual Segmentation Variability (Ground Truth) | Acceptable inter-rater and intra-rater variability for manual segmentations. | Evaluated via DSC in inter-rater and intra-rater variability studies. This study established the ground truth. The test results demonstrated that Constellation performs to its intended use and does not introduce new questions of safety or effectiveness. The device is claimed to be "as safe, as effective, and performs as well as the predicate devices." |
| Software Verification Testing | Passing results on all applicable unit, integration, and requirements testing. | Demonstrated passing results on all applicable unit, integration, and requirements testing according to "General Principles of Software Validation, Guidance for Industry and FDA Staff". |
| Software Usability Testing | Passing results for intended use. | Demonstrated passing results in the applied usability testing according to "Applying Human Factors and Usability Engineering to Medical Devices". |
2. Sample Size for the Test Set and Data Provenance
The provided document does not specify the sample size used for the test set.
The data provenance is also not explicitly stated beyond mentioning that the device "utilizes DICOM MR images gathered on a GE MR450W that encompass the whole-body" and is intended for "healthy adult patients." Whether the data was retrospective or prospective, or the country of origin, is not mentioned.
3. Number of Experts and Qualifications for Ground Truth
The document specifies:
- Number of experts: 3 (for Liver Volume of Interest (VOI) placement evaluation)
- Qualifications of experts: Radiologists. No further details (e.g., years of experience) are provided.
For manual segmentations (used to establish ground truth), it mentions "inter-rater and intra-rater variability studies," implying multiple raters, but does not specify the number or qualifications of these raters.
4. Adjudication Method for the Test Set
For the Liver Volume of Interest (VOI) placement evaluation, the adjudication method was majority voting of three radiologists.
For other segmentation accuracy evaluations and manual segmentations, the document does not explicitly state an adjudication method (like 2+1 or 3+1), but only mentions "inter-rater and intra-rater variability studies."
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
The document does not indicate that a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was done to measure the improvement of human readers with AI assistance versus without AI assistance. The study focuses on the standalone performance and repeatability of the device, as well as the establishment of its ground truth.
6. Standalone Performance Study
Yes, a standalone performance study was done. The document describes several performance tests, including:
- Segmentation accuracy of various anatomical regions (lungs, liver, spleen, kidneys, muscle, fat, brain structures) using DSC, mean percent absolute difference, and Pearson's correlation coefficient.
- Liver VOI placement evaluation.
- Device repeatability and test-retest measurement repeatability.
These evaluations were conducted on the algorithm directly, without a human-in-the-loop component being assessed as part of the primary performance metrics for the device itself. The device is positioned as a "clinical decision support tool," with the clinician retaining ultimate responsibility.
7. Type of Ground Truth Used
The ground truth for segmentation accuracy was established through manual segmentations by experts. This is indicated by: "Manual segmentations are established as ground-truth via inter-rater and intra-rater variability studies."
For Liver VOI placement, ground truth was implicitly established by the majority voting of three radiologists.
8. Sample Size for the Training Set
The document does not provide any information regarding the sample size for the training set.
9. How the Ground Truth for the Training Set Was Established
The document does not provide any information on how the ground truth for the training set was established. It only describes the establishment of ground truth for the test set through manual segmentations and expert review.
Ask a specific question about this device
Page 1 of 1