(28 days)
STAGE is a post-processing software medical device intended for use in the visualization of the brain. STAGE analyzes input data from MR imaging systems. STAGE utilizes magnitude and phase data acquired with specific parameters to generate enhanced Tl weighted images, susceptibility weighted imaging (SWI) images, susceptibility weighted image map (SWIM) images, pseudo-SWIM (pSWIM) images, modified pSWIM (mpSWIM) images, true SWI (tSWI) images, MR angiography (MRA) images, simulated dual-inversion recovery (DIR) images, and maps of TI, R2*, and proton density (PD).
When interpreted by a trained physician, STAGE images may provide information useful in determining diagnosis.
STAGE is indicated for brain imaging only and should always be used in combination with at least one other conventional MR acquisition ( e.g., T2 FLAIR).
STAGE works as a comprehensive brain imaging post-processing solution. The STAGE system consists of a client supplied dedicated computer with an ethernet connection to the client's existing local network. The STAGE software will operate within a virtual machine environment (virtual STAGE module) on this dedicated computer. The computer receives DICOM data from a specific MRI 3D GRE scan protocol (i.e., the STAGE protocol) and then outputs back numerous DICOM datasets with different types of contrast to the PACS server. The data transfer is initiated by the user's current DICOM viewing software. STAGE has been modified from the predicate to include CROWN, a white noise filtering algorithm intended to improve specific STAGE outputs.
The provided document is a 510(k) premarket notification for SpinTech, Inc.'s "STAGE" device. It outlines the device description, indications for use, comparison to a predicate device, and a summary of non-clinical testing. However, it explicitly states that clinical testing was not necessary to demonstrate substantial equivalence, and therefore, does not contain information about a study proving the device meets specific acceptance criteria based on human reader performance, expert-established ground truth, or MRMC studies.
Here's a breakdown of the information available based on your request, highlighting the missing elements:
1. A table of acceptance criteria and the reported device performance
The document does not provide a specific table of quantitative acceptance criteria and corresponding reported device performance metrics in the format requested (e.g., sensitivity, specificity, AUC values).
It states: "All predefined acceptance criteria for the performance testing were met. The results from the performance testing executed on STAGE produced results consistently according to its intended use." This is a general statement that the device passed internal testing, but no specific metrics are quantified or presented.
2. Sample size used for the test set and the data provenance (e.g. country of origin of the data, retrospective or prospective)
This information is not provided as there was no clinical testing. The non-clinical testing focused on software verification and validation.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g. radiologist with 10 years of experience)
This information is not provided as there was no clinical testing involving expert-established ground truth for a test set.
4. Adjudication method (e.g. 2+1, 3+1, none) for the test set
This information is not provided as there was no clinical testing (and thus no test set adjudicated by experts).
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
No, a multi-reader multi-case (MRMC) comparative effectiveness study was not done as explicitly stated, "Clinical testing was not necessary to demonstrate substantial equivalence of STAGE to the predicate device." Therefore, no effect size of human reader improvement with or without AI assistance is reported.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done
The document states: "STAGE was tested in accordance with SpinTech's verification and validation procedures," and "All predefined acceptance criteria for the performance testing were met." This implies standalone testing of the algorithm's performance against internal acceptance criteria (e.g., accuracy of calculations, image generation quality). However, specific metrics (like quantitative measures of accuracy, precision for the generated images or maps) for this standalone performance are not detailed. The focus is on the device generating outputs according to its intended methodology and not on diagnostic accuracy in a clinical context without human interpretation. The device is a "post-processing software medical device intended for use in the visualization of the brain" and "When interpreted by a trained physician, STAGE images may provide information useful in determining diagnosis." This indicates it's designed to be used with human interpretation, not as a standalone diagnostic tool.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc)
For the non-clinical performance testing, the "ground truth" would likely be based on established computational models, mathematical correctness of transformations, or comparison to reference data generated by known methods, rather than clinical ground truth like pathology or expert consensus. Specific details on the type of ground truth used for the verification and validation are not provided.
8. The sample size for the training set
This information is not provided. The document describes the device as a software that analyzes input data to generate various images and maps. While some methodologies might involve "fitting" (e.g., lease squares fitting), it's not explicitly stated if a machine learning model requiring a distinct "training set" was utilized in a conventional sense for the image generation process, beyond the CROWN filtering algorithm being "intended to improve specific STAGE outputs." The document details the methodologies as unchanged from the predicate, and focuses on the generation of images/maps based on specific signal processing and mathematical transformations, rather than a machine learning model that would require a large training dataset for learning.
9. How the ground truth for the training set was established
This information is not provided, as details about a distinct training set for a machine learning model are absent.
In summary, this 510(k) submission primarily focuses on demonstrating substantial equivalence to a predicate device through non-clinical software verification and validation, without the need for clinical studies involving human observers or detailed performance metrics against clinical ground truth.
§ 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.