(118 days)
Densitas densityai™ is a software application intended for use with compatible full field digital mammography and digital breast tomosynthesis systems. Densityai™ provides an ACR BI-RADS Atlas 5th Edition breast density category to aid interpreting physicians in the assessment of breast tissue composition. Densitas densityai™ produces adjunctive information. It is not a diagnostic aid.
Densitas densityai™ is a standalone software application that automatically analyzes "for presentation" data from digital breast x-ray systems, including digital breast tomosynthesis exams, to assess breast tissue composition. The software processes the data according to proprietary algorithms and generates a Breast Density Grade in accordance with the American College of Radiology's Breast Imaging Reporting and Data System (BI-RADS) 5th edition density classification scale. Densitas densityai™ data output is packaged for viewing on a mammography workstation or PACS as a DICOM mammography Structured Report or Secondary Capture. Output may also be transmitted to a RIS. Densitas densityai™ reports are configured to provide the following data based on the BI-RADS 5th edition breast density classification grade: For each patient: DENSITAS breast density grade (BDG).
Here's a summary of the acceptance criteria and study details for the densitas densityai™ device, based on the provided FDA 510(k) summary:
1. Acceptance Criteria and Reported Device Performance
The provided document does not explicitly state pre-defined acceptance criteria (e.g., minimum accuracy percentages). Instead, it presents the device's performance metrics, implying that these results were deemed acceptable for substantial equivalence. The performance is assessed by comparing the device's breast density classifications against expert radiologist consensus.
Table of Performance for densitas densityai™ (all scan types)
| Category Type | Performance Metric | Value |
|---|---|---|
| 4x4 Confusion Matrix (A,B,C,D Categories) | ||
| Overall Kappa | Kappa statistic (agreement with radiologist consensus) | 0.87 (95% CI: 0.87, 0.87) |
| Accuracy (Category A) | Device accuracy for Category A (Fatty) | 78% (72 correct out of 92 total A classifications by radiologists) |
| Accuracy (Category B) | Device accuracy for Category B | 76% (225 correct out of 295 total B classifications by radiologists) |
| Accuracy (Category C) | Device accuracy for Category C | 83% (262 correct out of 319 total C classifications by radiologists) |
| Accuracy (Category D) | Device accuracy for Category D (Extremely Dense) | 89% (84 correct out of 94 total D classifications by radiologists) |
| Grouped Confusion Matrix (Fatty A/B vs. Dense C/D) | ||
| Overall Kappa | Kappa statistic (agreement with radiologist consensus) | 0.84 (95% CI: 0.8, 0.88) |
| Accuracy (Fatty A/B) | Device accuracy for Fatty (A,B) categories | 88% (340 correct out of 387 total Fatty classifications by radiologists) |
| Accuracy (Dense C/D) | Device accuracy for Dense (C,D) categories | 96% (393 correct out of 409 total Dense classifications by radiologists) |
Additionally, other performance aspects were assessed through internal testing:
- Reliability: Measured by Pearson's Correlation Coefficient for percent density, dense breast area, and total area measurements between left and right breasts.
- Accuracy: Validated for percent density and total area measurements.
- Reproducibility: Measured by Pearson's Correlation Coefficient by running the algorithm twice over the same images.
- Inverse Relationship with Age: Pearson's Correlation Coefficient between percent density and age.
2. Sample Size and Data Provenance
- Test Set Sample Size: n=796 cases
- Data Provenance: Not explicitly stated regarding country of origin or whether it was retrospective/prospective. The description says "a large dataset that spanned all compatible scanner types," which implies diversity, but specifics are missing.
3. Number of Experts and Qualifications for Ground Truth
- Number of Experts: Four expert radiologists.
- Qualifications: "Expert radiologists" is stated, but no specific experience (e.g., years of experience, subspecialty certification) is detailed in the provided document.
4. Adjudication Method
- The ground truth was established by a "consensus assessment of four expert radiologists' independent readings." This implies an adjudication method where their individual readings were combined to form a single reference standard, but the specific consensus rule (e.g., majority vote, specific tie-breaking) is not described.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
- No MRMC comparative effectiveness study was explicitly mentioned. The study focused on the standalone performance of the AI device compared to expert consensus, not on how human readers' performance improved with AI assistance.
6. Standalone Performance
- Yes, a standalone performance study was done. The results presented in the tables (Kappa, Accuracy) directly assess the algorithm's performance in categorizing breast density against the expert radiologist consensus without human intervention.
7. Type of Ground Truth Used
- Expert Consensus: The ground truth for the validation study was established through "a consensus assessment of four expert radiologists' independent readings of overall breast composition." This falls under expert consensus.
8. Sample Size for the Training Set
- The document does not specify the sample size for the training set. It only describes the validation testing.
9. How the Ground Truth for the Training Set Was Established
- The document does not provide details on how the ground truth for the training set was established. Since the training set size isn't mentioned, neither are the methods for its ground truth. The information provided focuses solely on the validation/test set.
§ 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).