(46 days)
Quantib ND is a non-invasive medical imaging processing application that is intended for automatic labeling, visualization, and volumetric quantification of segmentable brain structures from a set of magnetic resonance (MR) images. The Quantib ND output consists of segmentations, visualizations and volumetric measurements of brain structures and white matter hyperintensities. Volumetric measurements may be compared to reference centile data. It is intended to provide the trained medical professional with complementary information and assessment of MR brain images and to aid the trained medical professional in quantitative reporting.
Quantib ND is an extension for Quantib Al Node software platform. It is intended for automatic labeling, visualization, and volumetric quantification of identifiable brain structures from magnetic resonance images (a 3D T1-weighted MR image for brain structure segmentation, with an additional T2-weighted FLAIR MR image for white matter hyperintensities (WMH) segmentation). The segmentation system relies on a number of atlases each consisting of a 3D T1-weighted MR image and a label map dividing the MR image into different tissue segments. Quantib ND provides quantitative information on both the absolute and relative volume of the segmented regions. The automatic WMH segmentation is to be reviewed and if necessary, edited by the user before validation of the segmentation, after which volumetric information is accessible. Quantib ND consists of 4 workflows: for both segmentation and quantification of brain structures as well as of white matter hyperintensities is there a single time-point analysis workflow, and a longitudinal workflow, which provides longitudinal analysis of images of two or more time-points. Quantib ND is intended to provide the trained medical professional with complementary information for the evaluation and assessment of MR brain images and to aid the radiology specialist in quantitative reporting.
Here's an analysis of the acceptance criteria and study proving device performance for Quantib ND 2.0, based on the provided FDA 510(k) summary:
1. Table of Acceptance Criteria and Reported Device Performance
The acceptance criteria for Quantib ND 2.0 are implicitly defined by the performance metrics (Dice index and Absolute difference of relative volumes) that demonstrate its substantial equivalence to the predicate device. While explicit "acceptance thresholds" are not stated, the reported performance is presented as proof of meeting the required level of accuracy for brain structure and white matter hyperintensity (WMH) segmentation.
| Brain Structure / Metric | Acceptance Criteria (Implicit) | Reported Device Performance (Mean ± Std. Dev.) |
|---|---|---|
| Brain Tissue (Dataset A) | ||
| Dice index | High overlap with manual | 0.96 ± 0.01 |
| Abs. diff. of rel. vol. | Low difference | 1.63 ± 1.06 pp |
| CSF (Dataset A) | ||
| Dice index | High overlap with manual | 0.78 ± 0.05 |
| Abs. diff. of rel. vol. | Low difference | 1.67 ± 1.06 pp |
| ICV (Dataset A) | ||
| Dice index | High overlap with manual | 0.98 ± 0.00 |
| Abs. diff. of rel. vol. | N/A | - |
| Hippocampus Total (Dataset B) | ||
| Dice index | High overlap with manual | 0.84 ± 0.03 |
| Abs. diff. of rel. vol. | Low difference | 0.03 ± 0.02 pp |
| Hippocampus Right (Dataset B) | ||
| Dice index | High overlap with manual | 0.84 ± 0.03 |
| Abs. diff. of rel. vol. | Low difference | 0.01 ± 0.01 pp |
| Hippocampus Left (Dataset B) | ||
| Dice index | High overlap with manual | 0.84 ± 0.04 |
| Abs. diff. of rel. vol. | Low difference | 0.01 ± 0.01 pp |
| Frontal Lobe Total (Dataset C) | ||
| Dice index | High overlap with manual | 0.95 ± 0.01 |
| Abs. diff. of rel. vol. | Low difference | 1.21 ± 1.22 pp |
| White Matter Hyperintensities (WMH) | ||
| Dice overlap | High overlap with manual | 0.61 ± 0.13 (over all cases) |
| Abs. diff. of rel. vol. | Low difference | 0.2 ± 0.2 pp (over 38 cases without CE) |
(Note: Only a subset of the detailed lobe data is included in the table for brevity. The full details are in the provided text.)
2. Sample Size Used for the Test Set and Data Provenance
- Brain Tissue, CSF, ICV (Dataset A): 33 T1w MR images.
- Hippocampus (Dataset B): 89 T1w MR images.
- Lobes (Dataset C): 13 T1w MR images.
- White Matter Hyperintensities: 45 3D T1w images (7 contrast-enhanced, all with corresponding T2w FLAIR images).
Data Provenance:
The document states that the test sets were "carefully selected to include data from multiple vendors and a series of representative scan settings." However, it does not specify the country of origin of the data or whether the data was retrospective or prospective.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts
The document states that the ground truth was established by "manual segmentations." It does not specify the number of experts involved in creating these manual segmentations or their qualifications (e.g., radiologist with X years of experience).
4. Adjudication Method for the Test Set
The document does not explicitly mention any adjudication method for the test set's ground truth beyond "manual segmentations," which implies that a single set of expert-generated segmentations served as the reference. There is no mention of 2+1, 3+1, or other consensus methods.
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done
No, the document does not describe an MRMC comparative effectiveness study where human readers' performance with and without AI assistance was evaluated. The study focuses solely on the performance of the algorithm against manual segmentations (ground truth). The device is intended to "aid the trained medical professional," but no study on this human-AI interaction is presented.
6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) Was Done
Yes, the provided performance data (Dice index and Absolute difference of relative volumes) explicitly represents the standalone performance of the Quantib ND 2.0 algorithm by comparing its automatic segmentations and volumetric measurements directly against expert manual segmentations (ground truth).
7. The Type of Ground Truth Used
The ground truth used for performance evaluation was expert manual segmentation.
- For brain structures, manual segmentations were performed on selected slices (Dataset A) or all slices (Datasets B and C) of T1w MR images.
- For White Matter Hyperintensities, manual segmentations were performed on T2w FLAIR images.
This type of ground truth is a form of expert consensus/reference standard, as it relies on human experts to delineate structures.
8. The Sample Size for the Training Set
The document does not specify the sample size used for the training set. It only describes the test sets used for validation.
9. How the Ground Truth for the Training Set Was Established
The document does not describe how the ground truth for the training set was established, as the details provided are limited to the performance evaluation of the final algorithm on the test sets.
§ 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).