BU-CAD
K210670TaiHao Medical Inc. · cleared 2021-12-21 · product code QDQ · Radiology
Premarket evidence — what FDA accepted
source quote (p.5)
“BU-CAD developed by TaiHao Medical Inc. is a software system designed to assist users in analyzing breast ultrasound images including identification of regions suspicious for breast cancer and assessment of their malignancy.”
source quote (p.9)
“Both BU-CAD and TransparaTM are intended to identify regions suspicious for breast cancer and provide computer analytics that are then synthesized by an artificial intelligence algorithm into a single value.”
source quote (p.20)
“The testing dataset was not used for training of BU-CAD algorithms.”
source quote (p.19)
“Draft Guidance for Industry and Food and Drug Administration Staff - Content of Premarket Submissions for Management of Cybersecurity in Medical Devices.”
Validation studies (2)
Reader study (MRMC)
n=628 cases · 2 site(s)
endpoints: AUC of location-specific ROC; sensitivity; specificity; PPV; NPV; BI-RADS descriptors determination
Standalone
n=1,139 cases
endpoints: lesion identification accuracy; AUC_LROC; sensitivity; specificity; PPV; NPV; robustness of CADx algorithm
Reported performance (10 observations)
source quote (p.14)
“Sensitivity: 0.9353 (0.9050, 0.9655)”
source quote (p.14)
“Specificity: 0.3611 (0.3124, 0.4098)”
source quote (p.14)
“Aided: 0.8160 (0.7862, 0.8458)”
source quote (p.14)
“NPV (unadjusted): 0.8945 (0.8456, 0.9434)”
source quote (p.14)
“PPV (unadjusted): 0.5056 (0.4607, 0.5505)”
source quote (p.18)
“Sensitivity (%): 88.33 (439/497) (0.8551, 0.9115)”
source quote (p.18)
“Specificity (%): 57.94 (372/642) (0.5413, 0.6176)”
source quote (p.18)
“PPV (%) [unadjusted]: 61.92 (439/709) (0.5834, 0.6549)”
source quote (p.18)
“NPV (%) [unadjusted]: 86.51 (372/430) (0.8328, 0.8974)”
source quote (p.16)
“The accuracy of the lesion identification algorithm was 93.24% (1062/1139).”
Each value carries its own analysis unit and task — never compare or pool across devices. Source: 510(k) summary PDF.
Predicate network
Postmarket — what happened after clearance
Recall and MAUDE counts are product-code-level (reports aren't reliably attributable to one device); a recall is shown as device-attributed only when the recall record itself lists this clearance number. Signals are descriptive observables with sources — never a judgment that the device is unsafe or drifting. Snapshot 2026-07-08.
Reimbursement — how devices like this got paid
Not yet tracked — no payment pathway indexed for this clearance (the reimbursement corpus is a growing seed set).
Applicable FDA guidance — what the submission is measured against
FDA guidance documents and guiding principles applicable to 510(k) AI/ML devices in the Radiology panel. A curated reference index, not legal or regulatory advice — each item states its own status, and a draft is never binding.
- Final guidanceRadiology-specific2022-09Computer-Assisted Detection Devices Applied to Radiology Images and Radiology Device Data - Premarket Notification [510(k)] Submissions
Radiology CADe/CADx · Software premarket content
Original July 2012; current database date reflects a Sept 2022 reissue. Governs CADe device 510(k) content.
- Final guidanceRadiology-specific2022-09Clinical Performance Assessment: Considerations for Computer-Assisted Detection Devices Applied to Radiology Images and Radiology Device Data in Premarket Notification (510(k)) Submissions
Radiology CADe/CADx
Original July 2012, revised 2020; current database date Sept 2022. Covers standalone and reader-study performance assessment for CADe.
- Final guidanceRadiology-specific2022-06Technical Performance Assessment of Quantitative Imaging in Radiological Device Premarket Submissions
Quantitative imaging · Radiology CADe/CADx
Final (June 2022). Relevant to devices outputting quantitative imaging measurements.
- Final guidance2026-01Clinical Decision Support Software
Clinical decision support · SaMD (general)
New final guidance issued Jan 2026, superseding the Sept 2022 version; narrows the device-CDS scope. Applies to software that informs clinical management.
- Final guidance2026-01General Wellness: Policy for Low Risk Devices
SaMD (general) · Clinical decision support
Revised final (Jan 2026); now addresses noninvasive products estimating physiologic parameters (SpO2, BP, glucose). Reshapes the device / non-device line for AI wellness features.
- Final guidance2025-09Computer Software Assurance for Production and Quality Management System Software
SaMD (general) · Postmarket
Final (Sept 2025). Covers software used in production/QMS (incl. ML development-pipeline tooling), superseding Section 6 of the 2002 GPSV — not device software functions themselves.
- Final guidance2025-06Cybersecurity in Medical Devices: Quality Management System Considerations and Content of Premarket Submissions
Cybersecurity · Software premarket content
Reissued June 2025 (retitled 'Quality Management System', was Sept 2023 'Quality System'); adds coverage of FD&C Act §524B cyber devices.
- Final guidance2024-12Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence-Enabled Device Software Functions
Predetermined Change Control Plan · AI/ML lifecycle · Software premarket content
Final (Dec 2024). Supersedes the April 2023 AI/ML PCCP draft.
- Final guidance2023-10Electronic Submission Template for Medical Device 510(k) Submissions
Software premarket content
eSTAR has been mandatory for 510(k)s since Oct 2023 — operationally unavoidable, though not AI-specific.
- Final guidance2023-08Off-The-Shelf Software Use in Medical Devices
Software premarket content · SaMD (general)
Final (Aug 2023). Applies when a device incorporates off-the-shelf software components (common in ML stacks).
- Final guidance2023-06Content of Premarket Submissions for Device Software Functions
Software premarket content · SaMD (general)
Final (June 2023); replaced the May 2005 'Software Contained in Medical Devices' guidance. Documentation level drives the software content of the submission.
- Final guidance2022-09Policy for Device Software Functions and Mobile Medical Applications
SaMD (general) · Clinical decision support
Current version Sept 2022. Frames which software functions FDA regulates as devices.
- Final guidance2021-10De Novo Classification Process (Evaluation of Automatic Class III Designation)
De Novo pathway
Final (Oct 2021), issued with the De Novo final rule. Most relevant to first-of-a-kind devices without a predicate (DEN-numbered clearances).
- Final guidance2016-12Postmarket Management of Cybersecurity in Medical Devices
Cybersecurity · Postmarket
- Final guidance2002-01General Principles of Software Validation
SaMD (general) · Software premarket content
Still active except Section 6 (superseded Sept 2025 by the Computer Software Assurance final guidance).
- Draft guidance2025-01Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations
AI/ML lifecycle · Software premarket content · Transparency
Draft as of July 2026 (published Jan 2025); finalization is on CDRH's FY2026 agenda but not yet published. Treat as FDA's stated direction, not a binding expectation.
- Draft guidance2024-08Predetermined Change Control Plans for Medical Devices
Predetermined Change Control Plan · Postmarket
Draft (Aug 2024) extending PCCPs beyond AI to all devices under FD&C §515C; not final as of July 2026.
- Guiding principles2024-06Transparency for Machine Learning-Enabled Medical Devices: Guiding Principles
Transparency · AI/ML lifecycle
- Guiding principles2023-10Predetermined Change Control Plans for Machine Learning-Enabled Medical Devices: Guiding Principles
Predetermined Change Control Plan · AI/ML lifecycle
FDA/Health Canada/MHRA joint principles (Oct 2023); companion to the GMLP and Transparency principles.
- Guiding principles2021-10Good Machine Learning Practice for Medical Device Development: Guiding Principles
AI/ML lifecycle · SaMD (general)
FDA/Health Canada/MHRA joint principles (Oct 2021). Foundational, not a binding guidance; IMDRF issued a related GMLP document Jan 2025.
Applicability is derived from the device's FDA advisory panel and pathway — cross-cutting guidances apply to every AI/ML device; panel-specific ones are flagged. Titles, dates, and links verified against fda.gov as of July 2026.