CEPHX3D

K252538

Orca Dental AI , Ltd. · cleared 2026-03-05 · product code QIH · Radiology

Premarket evidence — what FDA accepted

Device typesamd

source quote (p.8)

“CEPHX3D is a cloud-based software-as-a-service (SaaS) solution designed for the automated processing and visualization of dental and maxillofacial Cone Beam Computed Tomography (CBCT) data.”

Algorithm3D U-Net Convolutional Neural Networks (CNNs)

source quote (p.8)

“The device utilizes deep learning algorithms, specifically 3D U-Net Convolutional Neural Networks (CNNs) optimized for volumetric medical imaging, to segment specific anatomical structures from DICOM datasets.”

Adaptive (vs locked)FDA source did not state this

PCCPNo

Cybersecurity addressedYes

source quote (p.8)

“The software backend is built on the Spring Framework (Java) and hosted on Apache Tomcat (AWS Cloud), ensuring robust processing and data security.”

Validation studies (2)

Standalone

n=30 scans

endpoints: Dice Similarity Coefficient (DSC); Root Mean Square (RMS) surface error; 95th percentile Hausdorff Distance (HD95)

Retrospective clinical

n=53 cases

endpoints: Clinically Acceptable rating; Fleiss’ Kappa scores

Reported performance (6 observations)

diceas written: “Skeletal Bone Mean Dice Similarity Coefficient (DSC)”0.9827

source quote (p.13)

“Skeletal Bone: Achieved a mean Dice Similarity Coefficient (DSC) of 0.9827 and a mean Root Mean Square (RMS) surface error of 0.0623 mm.”

diceas written: “Dentition Mean Dice Similarity Coefficient (DSC)”0.9993

source quote (p.13)

“Dentition: Achieved a mean DSC of 0.9993 and a mean RMS of 0.0136 mm.”

diceas written: “Inferior Alveolar Nerve (IAN) Mean Dice Similarity Coefficient (DSC)”0.9861

source quote (p.13)

“Inferior Alveolar Nerve (IAN): Achieved a mean DSC of 0.9861 and a mean RMS of 0.1668 mm.”

diceas written: “Overall Global Mean DSC (Inter-Rater Reliability)”0.9905

source quote (p.13)

“The two independent experts achieved an Overall Global Mean DSC of 0.9905, confirming that the R-AS methodology provides a stable, reproducible, and objective anatomical baseline.”

agreement_kappaas written: “Fleiss' Kappa for bone and nerve models”1

source quote (p.14)

“Inter-rater reliability analysis confirmed the robustness of these results, showing Fleiss’ Kappa scores of 1.0 (Perfect Agreement) for bone and nerve models, and 0.80 (Strong Agreement) for dentition.”

agreement_kappaas written: “Fleiss' Kappa for dentition models”0.8

source quote (p.14)

“Inter-rater reliability analysis confirmed the robustness of these results, showing Fleiss’ Kappa scores of 1.0 (Perfect Agreement) for bone and nerve models, and 0.80 (Strong Agreement) for dentition.”

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

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).