(138 days)
GyriCalc is intended for automatic labeling, visualization, and quantification including volume, surface area and gyrification analysis (i.e., gyrification index) of segmentable brain structures from a set of MR images.
GyriCalc is intended to be used by qualified personnel and interpreted by a qualified clinician.
GyriCalc is not intended to be used for visualization or quantification of neurologic lesions.
GyriCalc is intended for children between 24 to 36 months of age.
GyriCalc is an automated imaging post-processing software medical device (SaMD) that provides automatic labeling, visualization, volumetric quantification, surface area, thickness and gyrification analysis of brain structures for children ages 24-36 months of age from a set of MR images and returns segmented images and morphometric reports. GyriCalc is a proprietary application which incorporates customized, state-of-the-art open-source software to perform image analysis and quantitative functionality.
The resulting output is provided as a PDF report with segmented color overlays and morphometric reports that can be displayed on commonly used Off The Shelf (OTS) PDF viewer. GyriCalc is not intended to be used for image review. The PDF report is the sole output of the device.
GyriCalc provides morphometric measurements based on 3D T1 MRI series. The output of the software includes volumes that have been annotated with color overlays, with each color representing a particular segmented region, and morphometric reports that provide measured volumes and other qualitative and quantitative data.
GyriCalc's processing architecture includes functionality that performs:
• Preprocessing
- Artifact correction (correct for various artifacts and distortions, such as motion, intensity inhomogeneity, and scanner-related differences),
- Skull-stripping to remove non-brain tissue,
- Bias field correction,
- Intensity normalization,
• Volumetric measurement,
• Surface area measurement,
• Morphological/morphometric analysis,
• Gyrification measurement and report generation.
Here's an analysis of the acceptance criteria and the study proving GyriCalc meets them, based on the provided FDA 510(k) Clearance Letter.
Acceptance Criteria and Device Performance for GyriCalc
1. Table of Acceptance Criteria and Reported Device Performance
| Metric | Region(s) Covered | Acceptance Criteria (Pre-specified) | Reported Device Performance (Mean Absolute Error) | Meets Criteria? |
|---|---|---|---|---|
| Dice's Coefficient (Segmentation Accuracy) | All regions | 0.92 - 0.99 (confidence range) | 0.92 - 0.99 (average of 0.95) | Yes |
| Volume Measurement Error | Total Cortex, Superior Frontal, Middle Frontal, Fusiform, Inferior Temporal, Lingual | Below 10% | 0.45% - 5.33% | Yes |
| Volume Measurement Error | Inferior Parietal Gyrus (Left) | Below 10% | 12.10% [8.44%, 15.77%] | No |
| Volume Measurement Error | Cuneus (Left & Right) | Below 10% | Left: 10.75% [8.50%, 13.00%], Right: 10.18% [7.79%, 12.57%] | No |
| Surface Area Measurement Error | Superior Frontal, Middle Frontal, Fusiform, Inferior Temporal, Lingual | Below 10% | 0.33% - 8.42% | Yes |
| Surface Area Measurement Error | Inferior Parietal Gyrus (Left) | Below 10% | 12.58% [8.72%, 16.43%] | No |
| Surface Area Measurement Error | Cuneus (Left & Right) | Below 10% | Left: 11.64% [8.75%, 14.54%], Right: 11.28% [8.51%, 14.05%] | No |
| Gyrification Index Error | All regions of interest | Below 10% | 0.04% - 0.89% | Yes |
Note: While some regions did not meet the stated pre-specified acceptance criteria for volume and surface area errors, the FDA still granted clearance, implying these deviations were considered acceptable in the overall context of safety and effectiveness, or that other factors mitigated the concern.
2. Sample Size and Data Provenance for the Test Set
- Sample Size: 82 T1 head MRIs from 82 patients.
- Data Provenance:
- Country of Origin: U.S. (54 patients) and Brazil (28 patients).
- Retrospective or Prospective: Retrospective. The imaging data was collected retrospectively from a population of anonymized patients with curated clinical records. The dataset represents a new, independent sampling of patients that were not involved in the development of the device.
3. Number of Experts and Qualifications for Ground Truth
- Number of Experts: A total of 6 experts were used, divided into two groups:
- Group 1 (32 cases): 2 U.S.-based neuroradiologists and 1 U.S.-based neuroimaging consultant PhD. (Specific years of experience are not mentioned, but their titles imply high qualification.)
- Group 2 (50 cases): 3 U.S.-based neuroradiologists. (Specific years of experience are not mentioned, but their titles imply high qualification.)
- Qualifications of Experts: Neuroradiologists and a Neuroimaging Consultant PhD. These are highly specialized medical professionals with expertise in interpreting and analyzing brain MRI images and neuroanatomy.
4. Adjudication Method for the Test Set
- Method: The segmentations of the 3 experts for each MRI were combined using the STAPLE method (Simultaneous Truth and Performance Level Estimation) to produce a single, consolidated segmentation. This method statistically estimates a consensus segmentation from multiple expert annotations, weighting each expert's contribution based on their estimated accuracy.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
- Was it done?: No, an MRMC comparative effectiveness study involving human readers assisting with or without AI was not reported. The study focused on the standalone performance of the AI algorithm against expert manual segmentations (ground truth). The document states, "The subject device (GyriCalc) and the predicate device (NeuroQuant) have the same automated quality control functions. Results must be reviewed by a qualified clinician." This indicates the device is intended as a support tool, implying a human-in-the-loop scenario, but the reported study does not compare human performance with and without the AI.
- Effect Size of Human Improvement with AI vs. Without AI Assistance: Not applicable, as this type of study was not conducted or reported.
6. Standalone (Algorithm Only) Performance Study
- Was it done?: Yes. The entire clinical performance assessment detailed in section 9.2 focuses on the standalone performance of the GyriCalc algorithm by comparing its output (segmentations and measurements) directly against expert-derived ground truth.
7. Type of Ground Truth Used
- Type: Expert Consensus Segmentation.
- For each brain MRI, three experts independently annotated and edited pre-loaded segmentations.
- These three expert segmentations were then combined using the STAPLE method to create a single, consolidated reference standard (ground truth).
- Reference measurements (volume, surface area, gyrification index) were then derived from this combined segmentation.
8. Sample Size for the Training Set
- Sample Size: The document does not specify the sample size used for the training set. It only describes the test set used for performance validation.
9. How Ground Truth for Training Set Was Established
- Method: The document does not explicitly state how the ground truth for the training set was established. It describes the ground truth establishment for the test set (expert consensus via STAPLE). For AI/deep learning models, training data often uses similar or less rigorous ground truth methods, or sometimes pseudo-labeling, but this information is not provided in the supplied text.
§ 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).