(135 days)
The SIGNA Premier system is a whole body magnetic resonance scanner designed to support high signal-to-noise ratio, and short scan times. It is indicated for use as a diagnostic imaging device to produce axial, sagittal, coronal, and oblique images, spectroscopic images, parametric maps, and/or spectra, dynamic images of the structures and/or functions of the entire body, including, but not limited to, head, neck, TMJ, spine, breast, heart, abdomen, pelvis, joints, prostate, blood vessels, and musculoskeletal regions of the body. Depending on the region of interest being imaged, contrast agents may be used.
The images produced by the SIGNA Premier system reflect the spatial distribution or molecular environment of nuclei exhibiting magnetic resonance. These images and/or spectra when interpreted by a trained physician vield information that may assist in diagnosis.
SIGNA Premier is a whole-body magnetic resonance scanner featuring a 3.0T superconducting magnet with a 70cm bore size. Major elements of the system include the magnet, gradient coils, body RF transmit coil, RF receive subsystem, patient support system (table), host computer, and system software. The system is compatible with a suite of RF receive coils, and is capable of using various pulse sequences, imaging techniques and reconstruction algorithms.
This submission is prompted by the introduction of a new software feature called AIR Recon DL onto the SIGNA Premier system. AIR Recon DL is a deep-learning based reconstruction technique designed to improve signal-to-noise ratio (SNR) and image sharpness. The feature also enables shorter scan times while preserving SNR and image sharpness.
The addition of the AIR Recon DL feature involved modifications to the SIGNA Premier system software. There were no changes related to AIR Recon DL to the system's hardware components.
The provided text describes the acceptance criteria and supporting studies for the SIGNA Premier system with the AIR Recon DL feature. However, it does not explicitly provide a table of acceptance criteria with reported device performance or all the specific details requested in question points 2 through 9.
Based on the available information, here's what can be extracted:
1. Table of Acceptance Criteria and Reported Device Performance:
| Acceptance Criteria (Inferred) | Reported Device Performance (AIR Recon DL) |
|---|---|
| Improvement in Signal-to-Noise Ratio (SNR) | Nonclinical: AIR Recon DL improves SNR. |
| Clinical: Objective measures of in vivo images confirmed AIR Recon DL improves SNR. Radiologists indicated a preference for AIR Recon DL images, implying improved SNR contributed to perceived image quality. | |
| Improvement in Image Sharpness | Nonclinical: AIR Recon DL improves image sharpness. |
| Clinical: Objective measures of in vivo images confirmed AIR Recon DL improves image sharpness. Radiologists indicated a preference for AIR Recon DL images, implying improved sharpness contributed to perceived image quality. | |
| Ability to Enable Shorter Scan Times (while maintaining SNR/sharpness) | Nonclinical: AIR Recon DL was able to maintain image SNR and did not sacrifice sharpness for images acquired with a reduced scan time. |
| Clinical: Comparisons were made between AIR Recon DL images from shorter scan time acquisitions and images without AIR Recon DL taken with longer scan times, with results confirming equivalent or better image quality for AIR Recon DL images. | |
| Low Contrast Detectability | Nonclinical: Maintained (did not negatively impact). |
| Noise Spectral Content | Nonclinical: Minimal impacts to. |
| Bias in Quantitative Measurements (based on signal intensity) | Nonclinical: No significant bias identified. |
| Appearance of Motion Artifacts | Nonclinical: Minimal impacts to (did not negatively impact). |
| Legibility of Clinically Relevant Structures | Clinical: Reader evaluation confirmed that AIR Recon DL provides images with equivalent or better image quality in terms of the legibility of clinically relevant structures. |
| Lesion Conspicuity | Clinical: Maintained with AIR Recon DL. |
| Overall Clinical Preference | Clinical: Radiologists reading the images indicated a preference for the AIR Recon DL images. Radiologists preferred the AIR Recon DL images for clinical use, even for samples with exogenous contrast and various pathologies. |
| Safety and Performance | Overall Conclusion: The nonclinical and clinical testing did not identify any new hazards, adverse effects, or safety or performance concerns that are significantly different from those associated with MR imaging in general. The device is at least as safe and effective as the predicate. |
2. Sample size used for the test set and the data provenance:
- Test Set Sample Size: The document mentions "objective measures of in vivo images" and a "reader evaluation study" on "images acquired across a variety of pulse sequences and anatomies," and "sample images from clinically indicated scans." However, the specific number of images or patient cases used for these test sets is not provided.
- Data Provenance: The document states "in vivo images" and "clinically indicated scans." This implies retrospective clinical data, but the country of origin is not specified.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- Number of Experts: The document states "Radiologists were asked to perform blinded reads" and "Radiologists were asked to rate the images." The specific number of radiologists/experts involved is not provided.
- Qualifications of Experts: The experts are identified as "Radiologists," but their specific qualifications (e.g., years of experience, subspecialty) are not provided.
4. Adjudication method for the test set:
- The document implies that radiologists provided ratings and comments, and the results were aggregated to conclude on preference and image quality. However, a formal adjudication method like "2+1" or "3+1" to establish a consensus ground truth among multiple readers is not explicitly stated or described. The reads were "blinded," but it doesn't detail how discrepancies were resolved or if there was a consensus process.
5. If a multi-reader multi-case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance:
- MRMC Study: A "reader evaluation study" was performed where "Radiologists were asked to perform blinded reads of both AIR Recon DL images and images without AIR Recon DL." This indicates a comparative reading study was conducted. Also, "Comparisons were also made between AIR Recon DL images from shorter scan time acquisitions and images without AIR Recon DL taken with longer scan times."
- Effect Size: The document states that the results "confirmed that the AIR Recon DL feature provides images with equivalent or better image quality in terms of the legibility of clinically relevant structures." It also notes "the radiologists reading the images also indicated a preference for the AIR Recon DL images." However, a specific quantifiable effect size measuring how much human readers improve (e.g., in terms of diagnostic accuracy, confidence, or reading time) with AI assistance compared to without it is not provided. The improvement is described qualitatively (equivalent or better image quality, preference).
6. If a standalone (i.e., algorithm only without human-in-the-loop performance) was done:
- The nonclinical testing on a "digital reference object and phantom imaging" evaluated the algorithm's impact on image quality metrics (SNR, sharpness, low contrast detectability, noise spectral content, bias, motion artifacts). This constitutes a standalone performance assessment of the algorithm's effects on image characteristics.
7. The type of ground truth used:
- For nonclinical testing (phantoms), the ground truth is known physical properties or measurements of the reference object/phantom.
- For clinical testing, the ground truth is established through expert consensus/ratings by radiologists. The document refers to "legibility of clinically relevant structures" and "lesion conspicuity" being maintained, relying on expert interpretation rather than pathology or long-term outcomes data.
8. The sample size for the training set:
- The document describes the AIR Recon DL feature as "a deep-learning based reconstruction technique." However, it does not provide any information regarding the sample size of the training set used to develop this deep learning model.
9. How the ground truth for the training set was established:
- As the document does not provide information about the training set (sample size or data), it does not describe how the ground truth for the training set was established.
§ 892.1000 Magnetic resonance diagnostic device.
(a)
Identification. A magnetic resonance diagnostic device is intended for general diagnostic use to present images which reflect the spatial distribution and/or magnetic resonance spectra which reflect frequency and distribution of nuclei exhibiting nuclear magnetic resonance. Other physical parameters derived from the images and/or spectra may also be produced. The device includes hydrogen-1 (proton) imaging, sodium-23 imaging, hydrogen-1 spectroscopy, phosphorus-31 spectroscopy, and chemical shift imaging (preserving simultaneous frequency and spatial information).(b)
Classification. Class II (special controls). A magnetic resonance imaging disposable kit intended for use with a magnetic resonance diagnostic device only is exempt from the premarket notification procedures in subpart E of part 807 of this chapter subject to the limitations in § 892.9.