MammoScreen® (4)

K243679

Therapixel · cleared 2025-07-03 · product code QDQ · Radiology

Premarket evidence — what FDA accepted

Device typesamd

source quote (p.8)

“MammoScreen is a software-only device.”

Algorithmdeep learning modules for the detection of suspicious calcifications and soft tissue lesions

source quote (p.8)

“The system includes ‘deep learning' modules for the detection of suspicious calcifications and soft tissue lesions. These modules are trained with very large databases of biopsy-proven examples of breast cancer and normal tissue.”

Adaptive (vs locked)FDA source did not state this

PCCPYes

source quote (p.10)

“MammoScreen is powered by machine-learning neural architecture. Therapixel will make future algorithm improvements under a PCCP. The plan describes future modifications and assesses their impact, and a modification protocol details how data management, re-training, performance evaluation and update procedures will be handled.”

Cybersecurity addressedFDA source did not state this

Validation studies (2)

Retrospective clinical

n=1,475 patients

endpoints: Non-inferiority in standalone cancer detection performance compared to the previous version of MammoScreen

standards: IEC 62304:2006/A1:2016- Medical device software - Software life-cycle processes, IEC 62366-1:2015+AMD1:2020- Medical devices - Application of usability engineering to medical devices.

Reader study (MRMC)

sample size not stated

endpoints: determine whether the radiologist's performance when using MammoScreen is superior to unaided radiologist performance for interpretation of mammograms.

Reported performance (3 observations)

aurocas written: “auc”0.894CI 0.870, 0.919

source quote (p.9)

“AUC at the mammogram level MS4: 0.894 (0.870, 0.919), MS2: 0.867 (0.839, 0.896), Δ: 0.027 (0.002, 0.052), p<0.0001”

aurocas written: “AUC at the breast level”0.919CI 0.897, 0.941

source quote (p.9)

“AUC at the breast level: MS4: 0.919 (0.897, 0.941), MS2: 0.895 (0.871, 0.920), Δ: 0.023 (0.002, 0.045), p<0.0001”

aurocas written: “AUC LROC at the finding level”0.891CI 0.862, 0.921

source quote (p.9)

“AUC LROC at the finding level: MS4: 0.891 (0.862, 0.921), MS2: 0.837 (0.797, 0.877), Δ: 0.055 (0.032, 0.077), p<0.0001”

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

recalls in product code, 24mo

MAUDE reports in code, 12mo

—

vs code's own 3-yr baseline

drift signals on this device

re_clearance
The FDA AI/ML device list shows a newer 510(k) K243685 (decision 2025-08-22) from Therapixel for a matching device line ("MammoScreen BD") — a new clearance for the same line is a change event.
first seen 2026-07-08 · k_number:K243685

Recall and MAUDE counts are product-code-level (reports aren't reliably attributable to one device). 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).