(29 days)
VBrain is a software device intended to assist trained medical professionals, during their clinical workflows of radiation therapy treatment planning, by providing initial object contours of known (diagnosed) brain tumors and organs at risk in the brain (i.e., the region of interest, ROI) on axial T1 contrast-enhanced brain MRI images. VBrain is intended to be used on adult patients only.
VBrain uses an artificial intelligence algorithm (i.e., deep learning neural networks) to contour (segment) brain tumor and organs at risk (brain stem, eyes, optic nerves, optic chiasm) in the brain on MRI images for trained medical professionals' attention, which is meant for informational purposes only and not intended for replacing their current standard practice of manual contouring process. VBrain does not alter the original MRI image, nor does it intend to detect tumors for diagnosis. VBrain is intended only for generating Gross Tumor Volume (GTV) contours of brain metastases, meningiomas, and acoustic neuromas, and contours of organs at risk in the brain; it is not intended to be used with images of other brain tumors or other body parts. The user must know the tumor type when they use VBrain.
VBrain also contains the automatic image registration feature to register volumetric medical image data. (e.g., MR, CT). It allows rigid image registration to adjust the spatial position and orientation of two images.
Medical professionals must finalize (confirm or modify) the contours generated by VBrain, as necessary, using an external platform available at the facility that supports DICOM-RT viewing/editing functions, such as image visualization software and treatment planning system.
VBrain is a software application system intended for use in the contouring (segmentation) of brain MRI images and in the registration of multi-modality images. The device consists of a workflow management module and 3 algorithm modules, which are the tumor contouring algorithm module, OAR contouring algorithm module, and registration algorithm modules can work independently, and yet can be integrated with each other.
The tumor contouring (segmentation) algorithm module consists of image preprocessing, deep learning neural networks, and postprocessing components, and is intended to contour brain tumor on the axial T1 contrast-enhanced MR images. It utilizes deep learning neural networks to generate contours for the detected/diagnosed brain tumors and export the results as DICOM-RT objects (using the RT Structure Set ROI Contour attribute, RTSTRUCT).
The OAR contouring (segmentation) algorithm module consists of image preprocessing, deep learning neural networks, and postprocessing components, and is intended to contour organs at risk in the brain on the axial T1 contrast-enhanced MR images. It utilizes deep learning neural networks to generate contours for the organs at risk in the brain and export the results as DICOM-RT objects (using the RT Structure Set ROI Contour attribute, RTSTRUCT).
The registration algorithm module registers volumetric medical image data (e.g., MR, CT). It allows rigid image registration to adjust the spatial position and orientation of two images.
The workflow management module is configured to work on a PACS network. Upon user's request, it will pull patient scans from a PACS, and it will trigger a predefined workflow, in which different algorithm modules are executed to generate the DICOM output. The DICOM output of a workflow can be sent back to the PACS.
The provided text describes a 510(k) summary for VBrain and references previous 510(k) submissions (K203235 and K212116) for the predicate devices. However, the current document does not explicitly state the acceptance criteria and the study results for the current device. It only mentions that "The protocol, methods and acceptance criteria of software verification and validation testing used to evaluate the changes were not modified from those used in the predicate submission. The acceptance criteria and a summary of the results were provided for each test. VBrain passed all V&V testing, performance requirements and specifications are met."
To provide a complete answer, I would need access to the predicate submissions (K203235 and K212116) which presumably contain the detailed acceptance criteria and study results.
Based only on the provided text, here’s what can be inferred or stated:
1. A table of acceptance criteria and the reported device performance
The document states that "The acceptance criteria and a summary of the results were provided for each test. VBrain passed all V&V testing, performance requirements and specifications are met." However, the specific criteria and performance values are not detailed in this submission. This document highlights that the protocols, methods, and acceptance criteria were not modified from the predicate submissions, implying that the performance metrics from the predicate devices are applicable and the current device met those established benchmarks.
To present a table, I would need the specific metrics (e.g., Dice Similarity Coefficient, Hausdorff Distance, etc.) and their thresholds from the predicate summaries, which are not in the provided text.
2. Sample size used for the test set and the data provenance
This information is not provided in the current document. It retrospectively refers to the V&V testing from the predicate submissions.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts
This information is not provided in the current document. It retrospectively refers to the V&V testing from the predicate submissions.
4. Adjudication method for the test set
This information is not provided in the current document. It retrospectively refers to the V&V testing from the predicate submissions.
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
The document describes VBrain as a "software device intended to assist trained medical professionals... by providing initial object contours," and states that "Medical professionals must finalize (confirm or modify) the contours generated by VBrain". This indicates a human-in-the-loop workflow. However, the current document does not report on a direct MRMC comparative effectiveness study or the effect size of human reader improvement with AI assistance. This information, if available, would likely be in the predicate submissions.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done
The document implies standalone performance evaluation based on the statement that VBrain "generates contours for the detected/diagnosed brain tumors and exports the results as DICOM-RT objects." The V&V testing would have evaluated the accuracy of these generated contours against ground truth. The acceptance criteria for such standalone performance are referenced to the predicate submissions but are not explicitly listed here.
7. The type of ground truth used
This information is not explicitly provided in the current document. It retrospectively refers to the V&V testing from the predicate submissions. It is common for such devices to use expert consensus contours (often by radiologists or radiation oncologists) as the ground truth for segmentation accuracy, but this is not confirmed in the provided text.
8. The sample size for the training set
This information is not provided in the current document.
9. How the ground truth for the training set was established
This information is not provided in the current document.
In summary of what is available:
The current 510(k) submission for VBrain (K213628) primarily focuses on the substantial equivalence of the modified device to its two predicate devices (K203235 and K212116). It states that the "protocol, methods and acceptance criteria of software verification and validation testing used to evaluate the changes were not modified from those used in the predicate submission" and that "VBrain passed all V&V testing, performance requirements and specifications are met." Therefore, the detailed acceptance criteria and study particulars are implicitly relying on the documentation provided in the predicate submissions, which are not included in the provided text.
§ 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).