Search Results
Found 1 results
510(k) Data Aggregation
(493 days)
QIR Suite is intended to be used for viewing, post-processing, and quantitative evaluation of cardiovascular Magnetic Resonance (MR) images in a DICOM (Digital Imaging and Communication in Medicine) Standard format. The software has been validated for use on adult patients.
QIR Suite comprises QIR-MR for analysis of MR images. QIR-MR is composed of a viewer and analysis modules, and uses user inputs, standard algorithms, and/or automated deep learning detection algorithms.
QIR Suite support the following functionalities:
· Receive, store, transmit, post-process, display, and manipulate medical MR/CT images in the DICOM format (all transfer syntaxes supported including JPEG2000).
-
· Client/server functionalities to connect to a PACS (Picture Archiving and Communication System), to a HL7 server.
· Visualization of 2D and 2D + time of single or multiple MR datasets. -
· Segmentation of regions of interest.
-
· Measurement of distances and areas.
· Cardiac function MR analyses for the four chambers, including ejection assessment, local myocardial mass, diastolic function, thickness and thickening.
• 2D Flow studies.
Each module generates an automated report of the analysis. QIR Suite allows connection and storage of analyses on a PACS and on a HL7 server.
The software is not intended for use by patients, but rather by qualified medical professionals, experienced in examining and interpreting cardiovascular MR images to obtain diagnostic information as part of a comprehensive diagnostic decision-making process. OIR Suite cannot replace the diagnosis of a qualified practitioner and cannot be regarded as a sole medical point-of-view. The final diagnosis is the sole responsibility of the practitioner.
QIR Suite is a software for quantitative analyses of cardiovascular magnetic resonance images in the DICOM format. Analyses are performed using standardized and deep-learning algorithms. QIR Suite has been validated for adult patients. QIR Suite is intended to be used by qualified medical professionals, experienced in examining and evaluating cardiovascular MR images for the purpose of obtaining diagnostic information, as part of a comprehensive diagnostic decision-making process. QIR Suite cannot replace the diagnosis of a qualified practitioner and cannot be regarded as a sole medical point-of-view.
Acceptance Criteria and Device Performance Study for QIR Suite
1. Table of Acceptance Criteria and Reported Device Performance
The performance testing for QIR Suite focused on demonstrating substantial equivalence to predicate devices (Segment CMR and CVI42) by comparing quantitative measurements and evaluating deep learning algorithm performance. The acceptance criteria and reported device performance are summarized below:
| Feature/Parameter Tested | Acceptance Criteria | Reported Device Performance |
|---|---|---|
| Quantitative Parameters (Comparison with Predicate Devices) | ||
| Correlation Coefficient (R²) for all measurements (QIR Suite vs. Predicate) | R² > 0.95 | Systematically above 0.97, with an average correlation above 0.99 for all comparisons. Specifically, for cardiac function parameters, the minimum R² was 0.9792, and most were above 0.99, with an average of 0.9954. For 2D flow parameters, the minimum R² was 0.9590, and most were above 0.99, with an average of 0.9907. |
| Absolute Mean Difference for all measurements (QIR Suite vs. Predicate) | < 10% | Well under 10% for all parameters. Specifically: - Distance measurements: 0.8% variation (over 11 measurements) - Area measurements: 2.8% variation (over 10 measurements) - Cardiac function parameters: Well under 5% - 2D flow parameters: Under 10% |
| Deep Learning Algorithms | ||
| Dice Coefficient (against ground truth) | "Closed to previously published results" (no specific numerical threshold provided, but implied to be high due to the context of high agreement) | Mean score of 0.893 for AG algorithm Mean score of 0.888 for AG+ algorithm Mean score of 0.908 for Fast algorithm |
2. Sample Size Used for the Test Set and Data Provenance
- Sample Size for Test Set: The document does not specify a precise numerical sample size for the test set used for quantitative parameter comparison. It refers to "patient examinations" and "a dataset comprised of MR images from patients."
- Data Provenance: The data used for testing were from a "patients' database from the US and Europe," and "patients from Europe, the USA, and India."
- Retrospective/Prospective: The data appears to be retrospective, as it refers to "recorded" data and "patients' database."
3. Number of Experts Used to Establish Ground Truth for the Test Set and Qualifications
The document does not explicitly state the number of experts used to establish the ground truth for the test set or their specific qualifications (e.g., "radiologist with 10 years of experience").
For quantitative parameters, the "ground truth" was established by measurements performed using predicate devices (Segment CMR and CVI42), implying that the predicate devices' outputs serve as a reference assumed to be "true." For the deep learning algorithms, a "ground truth" was used, but the method of its establishment (e.g., expert consensus) is not detailed.
4. Adjudication Method for the Test Set
The document does not describe an explicit adjudication method (e.g., 2+1, 3+1) for establishing ground truth for the test set during performance testing. For quantitative parameters, the comparison was directly between QIR Suite and predicate device measurements. For deep learning algorithms, it was against a "ground truth," but the adjudication process for that ground truth is not specified.
5. Multi Reader Multi Case (MRMC) Comparative Effectiveness Study
There is no indication that a Multi Reader Multi Case (MRMC) comparative effectiveness study was done to assess how much human readers improve with AI vs without AI assistance. The study focuses on the standalone performance of the QIR Suite and its equivalence to predicate devices, rather than its impact on human reader performance.
6. Standalone Performance (Algorithm Only)
Yes, a standalone (algorithm only) performance study was done. The performance data section describes extensive testing where the output of each function of QIR Suite was compared against predicate devices or a defined "ground truth" (for deep learning algorithms). This indicates an evaluation of the algorithm's performance without a human-in-the-loop component.
7. Type of Ground Truth Used
- For quantitative parameters (e.g., distance, area, cardiac function, 2D flow), the "ground truth" was effectively established by the measurements obtained from legally marketed predicate devices (Segment CMR and CVI42).
- For deep learning algorithms, a "ground truth" was used, but the specific type (e.g., expert consensus, pathology) is not detailed. It is implied to be a reference standard against which the algorithm's segmentation performance was evaluated using the Dice coefficient.
8. Sample Size for the Training Set
The document does not specify the sample size for the training set used for the deep learning algorithms. It mentions "large testing datasets" for evaluating deep learning algorithms, but a separate size for the training set is not provided.
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 for the deep learning algorithms. It only mentions that the deep learning algorithms "were evaluated against a ground truth" during performance testing.
Ask a specific question about this device
Page 1 of 1