(102 days)
ASPIRE Bellus is intended to receive digital mammography images and to display these images on monitors for radiologists' review for diagnostic and screening purposes. To assist radiologists, ASPIRE Bellus provides functions such as image review, measurement, post-processing, film printing, displaying mammography CAD results, and image manipulation.
ASPIRE Bellus does not accept lossy compressed mammographic images, which should not be used for primary diagnostic interpretation. Display monitors connected to ASPIRE Bellus for diagnostic interpretation of mammographic images must be cleared for use in digital mammography. All images sent to or imported into the ASPIRE Bellus must conform to regulatory requirements. Image quality must conform to applicable quality guidelines.
FUJIFILM's Mammography Viewer SMV658 (V3.0) system receives mammography images directly from FUJIFILM's digital mammography acquisition systems using DICOM protocol and PACS via network. These images are displayed on mammography diagnostic monitors for doctors' review. The SMV658 system works as mammography workstation depending on its configuration (monitors) and license information (software key). The system provides visualization and image enhancement tools (such as image review, measurement, postprocessing, film printing, displaying mammography CAD results, and image manipulation) to assist the radiologists' review of mammography images for diagnostic and screening purposes.
Here's an analysis of the provided text regarding the acceptance criteria and study for the FUJIFILM Medical Systems U.S.A., Inc. SMV658 V3.0 (ASPIRE Bellus / Mammography Viewer SMV658) 510(k) submission.
Critical Note: The provided 510(k) summary is for a mammography viewer software (PACS), not an AI or CAD device. Therefore, the typical performance metrics (sensitivity, specificity, AUC) and study designs (MRMC, standalone) associated with AI diagnostic aids are not present because the device's function is image display and manipulation, not diagnostic interpretation. The "acceptance criteria" discussed are primarily related to software functionality and safety, not diagnostic accuracy.
Acceptance Criteria and Reported Device Performance
| Acceptance Criteria Category | Reported Device Performance | Comments from the Document |
|---|---|---|
| Software Functionality | Passed all tests | "Mammography Viewer SMV658 is tested successfully with reference to its Software Requirements Specification, as well as design verification and validation documents and Traceability Matrix document." "Test results showed that all tests successfully passed." |
| Segmentation Accuracy | Passed all tests | "Testing involved system level functionality test, segmentation accuracy test, measurement accuracy test, interfacing test, usability test, serviceability test, labeling test, as well as the test for risk mitigation method analyzed and implemented in the risk management process." |
| Measurement Accuracy | Passed all tests | (See above for Segmentation Accuracy) |
| Interfacing | Passed all tests | (See above for Segmentation Accuracy) |
| Usability | Passed all tests | (See above for Segmentation Accuracy) |
| Serviceability | Passed all tests | (See above for Segmentation Accuracy) |
| Labeling | Passed all tests | (See above for Segmentation Accuracy) |
| Risk Mitigation | Passed all tests | (See above for Segmentation Accuracy) |
| Bench Performance (with clinical images) | Achieved expected accuracy performance | "In addition, we conducted the bench performance testing using actual clinical images to help demonstrate that the proposed device achieved the expected accuracy performance." |
| Safety and Effectiveness | Deemed safe and effective and substantially equivalent to predicate | "Verification, validation, and testing activities establish the performance, functionality, and reliability characteristics of the Mammography Viewer SMV658 software, which is found to be safe and effective and substantially equivalent to the currently cleared predicate devices." |
Study Details (Based on the provided text)
-
Sample size used for the test set and the data provenance (e.g. country of origin of the data, retrospective or prospective):
- The document mentions "bench performance testing using actual clinical images." However, it does not specify the sample size of these clinical images.
- The provenance (country of origin, retrospective/prospective) of the clinical images used for bench performance testing is not mentioned in the provided text.
-
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):
- The document does not describe an expert-based ground truth establishment process for a diagnostic performance study. The "testing" mentioned is primarily focused on software functionality and its ability to display images and perform measurements as intended. For "bench performance testing using actual clinical images," it's implied that the accuracy of the software's display and measurement capabilities was assessed, but not against a "ground truth" of disease presence/absence established by experts for diagnostic purposes.
-
Adjudication method (e.g. 2+1, 3+1, none) for the test set:
- Not applicable / not mentioned. Since there is no expert-based ground truth establishment for diagnostic performance, adjudication methods for disagreements among experts are not relevant to the described testing.
-
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, an MRMC comparative effectiveness study was not done. The device is a mammography viewer, not an AI or CAD system intended to provide diagnostic interpretations or assist human readers in lesion detection/diagnosis. Therefore, the concept of human reader improvement with or without AI assistance is not relevant to this device's intended use and testing.
-
If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:
- No, a standalone diagnostic performance study was not done. This device is a viewer that supports radiologists; it is not an algorithm designed to make diagnostic interpretations or detections independently. Its "performance" relates to its functionality, display accuracy, and measurement capabilities, not standalone diagnostic accuracy. The bench testing with clinical images likely validated its ability to correctly display and measure features, not to diagnose.
-
The type of ground truth used (expert consensus, pathology, outcomes data, etc.):
- For the "bench performance testing using actual clinical images," the document does not explicitly state the type of "ground truth" used. However, given the device's function (viewer with measurement tools), the "ground truth" for these tests would likely involve:
- Reference measurements: For measurement accuracy, a known, precise measurement (e.g., from a calibrated tool or pre-adjudicated measurement on the image) would serve as ground truth to compare against the device's measurement function.
- Image integrity/display fidelity: For display-related performance, the ground truth would be the original image data and its accurate representation on the monitor.
- It is not using pathology, expert consensus on disease, or outcomes data, as those are typical for diagnostic AI devices, which this is not.
- For the "bench performance testing using actual clinical images," the document does not explicitly state the type of "ground truth" used. However, given the device's function (viewer with measurement tools), the "ground truth" for these tests would likely involve:
-
The sample size for the training set:
- Not applicable / not mentioned. This device is an image viewer with image processing and manipulation tools. It does not appear to be an AI or machine learning model that requires a "training set" in the conventional sense for diagnostic prediction. Its development likely involved standard software engineering practices and testing against specifications.
-
How the ground truth for the training set was established:
- Not applicable / not mentioned. As there's no indication of a training set for an AI/ML model, the establishment of ground truth for such a set is not discussed.
Summary of Device and Context:
It's crucial to understand that the FUJIFILM Medical Systems U.S.A., Inc. SMV658 V3.0 (ASPIRE Bellus / Mammography Viewer SMV658) is a PACS (Picture Archiving and Communications System) or a specialized workstation for viewing digital mammography images. Its primary functions are receiving, displaying, enhancing, and manipulating images, as well as displaying CAD results (generated by other cleared CAD systems, not the viewer itself).
Therefore, the "acceptance criteria" and "study" described in this 510(k) are focused on:
- Software verification and validation: Ensuring the software functions as designed, is reliable, and meets its specifications.
- Safety: Hazard analysis and mitigation.
- Substantial equivalence: Demonstrating that it performs similarly to other legally marketed mammography viewers without raising new safety or effectiveness concerns.
It is not a diagnostic AI device, and thus, its regulatory submission does not include the typical performance metrics or study designs (e.g., MRMC, standalone performance, sensitivity/specificity) associated with AI-based diagnostic aids.
§ 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).