(248 days)
cNeuro cMRI is intended for automatic labeling, quantification of segmentable brain structures from a set of MR images. The software is intended to automate the current manual process of identifying, labeling and quantifying the segmentable brain structures identified on MR images.
The intended user profile covers medical professionals who work with medical imaging. The intended operational environment is an office-like environment with a computer.
cNeuro cMRI is intended for automatic labeling, quantification of segmentable brain structures from a set of MR images. The software is intended to automate the current manual process of identifying, labeling and quantifying the segmentable brain structures identified on MR images.
As input, cNeuro cMRI uses T1-weighted (T1) and fluid-attenuated inversion recovery (FLAIR) DICOM MR images from a single time point. The T1 image is mandatory but the FLAIR image is optional. The user selects images through connection with a Picture Archiving and Communication System (PACS) or by selecting DICOM files from a folder. cNeuro cMRI displays the selected images together with information extracted from the DICOM headers.
lmage processing starts with a pre-processing stage with bias-field correction and brain extraction before the actual segmentation and calculation of MRI biomarkers begins. When the processing has completed, the user can review the images with brain segmentations displayed as an overlay. cNeuro cMRI presents computed biomarkers corresponding to volumes of structures and FLAIR white matter hyperintensities. The computed biomarkers are corrected for the subject's head size, gender and age and are compared to corresponding biomarkers from a healthy reference population using a statistical model.
Here's a detailed breakdown of the acceptance criteria and study information for the cNeuro cMRI device, based on the provided text:
1. Table of Acceptance Criteria and Reported Device Performance
The document explicitly states that "A literature review was performed to set relevant acceptance criteria for each type of experiment. All experiments passed the acceptance criteria." While the specific numerical acceptance criteria from the literature review are not detailed in the provided text, the reported device performance for key metrics is given.
| Metric | Acceptance Criteria (Not explicitly stated numerically in source, but "passed") | Reported Device Performance |
|---|---|---|
| Similarity Index (Dice Index) - Hippocampus | Value from literature review | 0.88 |
| Similarity Index (Dice Index) - Thalamus | Value from literature review | 0.91 |
| Similarity Index (Dice Index) - Whole Cortex | Value from literature review | 0.88 |
| Intraclass Correlation Coefficient (ICC) - Test-Retest Reproducibility (averaged over 133 structures) | Value from literature review | 0.96 |
| Correlation Coefficient - FLAIR White Matter Hyperintensities (vs. manually labeled) | Value from literature review | 0.97 |
2. Sample Size Used for the Test Set and Data Provenance
- Sample Size for Test Set: 1399 subjects in total.
- Data Provenance: The test data included "data from healthy subjects, and patients with neurodegenerative diseases such as Alzheimer's disease, mild cognitive impairment, fronto-temporal lobe degeneration, vascular dementia as well as Multiple Sclerosis patients." The country of origin is not specified, nor is 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 mentions "manually labeled ground truth data" and "manually labelled data" for white matter hyperintensities. However, it does not specify the number of experts used or their qualifications (e.g., radiologists with X years of experience).
4. Adjudication Method for the Test Set
The document does not specify an adjudication method (e.g., 2+1, 3+1, none) for the test set. It only mentions "manually labeled ground truth data."
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
A multi-reader multi-case (MRMC) comparative effectiveness study was not explicitly mentioned in the provided text. The study focused on validating the accuracy and reproducibility of the automated system against ground truth, not on human reader performance with or without AI assistance.
6. Standalone Performance Study
Yes, a standalone study was done. The performance metrics reported (Similarity Index, ICC, Correlation Coefficient) directly reflect the algorithm's performance in automatically segmenting and quantifying brain structures in comparison to "manually labeled ground truth data" or test-retest data, without direct human-in-the-loop interaction for the reported metrics. The "QC of segmentation results" and "reviewing biomarkers" in the workflow indicate a human review step, but the reported performance metrics are for the algorithmic output.
7. Type of Ground Truth Used
The ground truth used was expert consensus / manual labeling. Specifically, the document states:
- "In the accuracy experiments, cNeuro cMRI fully automated brain segmentation was compared to manually labeled ground truth data."
- "and the correlation coefficient between the computed FLAIR white matter hyperintensities and the manually labelled data was 0.97."
8. Sample Size for the Training Set
The document does not specify the sample size used for the training set. The 1399 subjects are mentioned in the context of "experiments," which typically implies testing or validation rather than training.
9. How the Ground Truth for the Training Set Was Established
The document does not provide information on how the ground truth for the training set was established.
§ 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).