(146 days)
VersaViewer is a medical diagnosis software supporting 2D, 3D and 4D medical images series for their processing and analysis through customizable layouts allowing multimodality review. It streamlines standard and advanced medical imaging analysis by providing a suite of measurements capabilities. It is designed for use by trained healthcare professionals and is intended to assist the physician in diagnosis, who is responsible for making all final patient management decisions.
VersaViewer is not intended for the displaying of digital mammography images for diagnosis.
VersaViewer is a software application for processing and analysis of 2D, 3D and 4D medical imaging data. The application provides adaptive layout to display selected series and, common radiology toolset to perform measurements. It aims to enable the review of medical imaging acquisitions for which a dedicated advanced visualization application is not required.
VersaViewer has the following functionalities:
- Reconstruct and display 2D, 3D and 4D medical images from multiple modalities.
- Display relevant series in an adaptive layout based on user selection.
- Access and dynamically load series of interest through embedded Series Selector.
- Allow to select different image rendering modes such as Volume Rendering, MIP (maximum intensity projection) /MinIP (minimum intensity projection) / Average, MPR (multiplanar reformation) and Oblique.
- Basic image review tools including paging, WW/WL adjustment and zoom. 3D volumes can be visualized in adjustable multi-oblique planes.
- Set of annotation, measurement, and segmentation tools.
- Dedicated panel collects findings as they are deposited on the images and enables user to manage them.
- Images and findings export options.
VersaViewer also includes One View as an optional feature.
One View provides reformatted views to assist radiologists in interpreting various types of spectral exams by projecting GSI (Gemstone Spectral Imaging) material decomposition images over monochromatic and color overlay.
Here's an analysis of the acceptance criteria and study detailed in the provided FDA 510(k) submission for VersaViewer, structured as requested:
Acceptance Criteria and Device Performance for VersaViewer
The provided 510(k) summary for VersaViewer indicates that the device's performance was evaluated through engineering bench testing for its newly introduced deep learning algorithm, specifically for automated segmentation. While explicit, numerically stated acceptance criteria and corresponding reported device performance values are not provided in the document, the conclusion states:
"The result of the algorithm validation showed that the algorithm successfully passed the defined acceptance criteria."
This implies that internal, pre-defined acceptance criteria were established and met. Without the actual criteria and performance metrics, a definitive table cannot be generated. However, based on the text, we can infer the nature of the evaluation.
Inferred Acceptance Criteria & Reported Performance:
| Acceptance Criteria (Inferred) | Reported Device Performance |
|---|---|
| Accuracy/Robustness of Automated Segmentation for six body parts (lung, liver, bone, aorta, heart, entire body) using the deep learning algorithm. | "The algorithm successfully passed the defined acceptance criteria." (Implies satisfactory accuracy and robustness as per internal thresholds.) |
| Successful Design Verification and Validation for all functionalities, including image rendering, manipulation, measurement, and export. | "VersaViewer has successfully completed the design control testing per GE HealthCare's quality system." |
| "All the testing and results did not raise new or different questions of safety and effectiveness other than those already associated with predicate devices." | |
| Compliance with DICOM Standards | "The proposed device complies with NEMA PS 3.1 - 3.20 (2023) Digital Imaging and Communications in Medicine (DICOM) Set (Radiology) standard." |
1. Sample Size for Test Set and Data Provenance
- Sample Size for Test Set: The document states "a database of retrospective CT exams." It does not specify the exact number of exams or cases in this database.
- Data Provenance: The data used was from "retrospective CT exams." The country of origin is not explicitly mentioned, but GE Medical Systems SCS (the submitter) is based in France, and GE Medical Systems, LLC (reference device manufacturer) is in the US, suggesting a multinational potential. Given it's a 510(k) for a global company, it's common for data to be sourced from multiple regions, but this is not confirmed here.
2. 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 or their qualifications for establishing ground truth for the test set of the deep learning algorithm. It only mentions that the deep learning algorithm's segmentation was compared "to legacy segmentation algorithms" during verification and validation. This suggests a comparison against existing, validated methods rather than a direct human expert ground truth adjudication for the algorithm's performance.
3. Adjudication Method for the Test Set
The document does not specify an adjudication method (e.g., 2+1, 3+1) for establishing ground truth for the deep learning algorithm's performance on the test set. The validation primarily involved comparison to "legacy segmentation algorithms," implying a technical comparison rather than an expert consensus process for newly established ground truth for each case.
4. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
No mention of a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was made in the provided document. Therefore, there is no information about the effect size of how much human readers improve with AI vs. without AI assistance. The VersaViewer is described as a medical diagnosis software with tools for processing and analysis, assisting the physician, but not primarily as an AI-assistance tool for diagnostic improvement (though its One View feature uses deep learning for segmentation).
5. Standalone Performance Study (Algorithm Only Without Human-in-the-Loop)
Yes, a standalone performance study was implicitly done for the deep learning algorithm. The document states:
"Engineering has performed bench testing for the newly introduced deep learning algorithm in the subject device for automated segmentation of six body parts as lung, liver, bone, aorta, heart and body (entire body) using a database of retrospective CT exams."
This describes the evaluation of the algorithm's performance in isolation.
6. Type of Ground Truth Used
For the deep learning segmentation algorithm, the ground truth was established by comparison to legacy segmentation algorithms. The text states: "...the deep learning algorithm employed in the subject device have been successfully verified and validated, through comparison to legacy segmentation algorithms."
7. Sample Size for the Training Set
The document does not provide the sample size used for the training set of the deep learning algorithm. It only refers to a "database of retrospective CT exams" used for bench testing of the algorithm, which is typically the test set or validation set.
8. 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. It only mentions the comparison to "legacy segmentation algorithms" for the verification and validation (i.e., testing) phase. The methods for annotating or establishing ground truth for the data used to train the deep learning model are not detailed.
§ 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).