Search Results

Segmentron Viewer is a software product intended for processing and manipulating maxillofacial radiographic images. Segmentron Viewer allows users to perform the following functions:

1. Viewing patient images (provides tools for image processing and viewing functions);
2. Reading and 3D visualization of CBCT images;
3. Generating editable 3D STL files (for educational purposes only).

The device is indicated for use by medical professionals (such as dentists and radiologists), in patients 14 years and older with permanent teeth.

Segmentron Viewer is a web application. It can be used in a network environment.

Segmentron Viewer is a semi-automated software as a medical device (SaMD) for dental image processing and management. The device's main function is to perform automated analysis of maxillofacial Cone Beam Computed Tomography (CBCT) images uploaded by the user, which consists of applying artificial neural network models (AI) to such images to obtain automatically generated 3D segmentations of teeth and anatomy. The user is able to edit these segmentations. The device also provides functions for enhancement and 3D visualization of the images. It additionally enables uploading, saving, and sharing CBCT images for the clinician's ease of use.

Segmentron Viewer identifies each tooth and tooth pulp present in the upper and the lower jaw (as shown on the input scan), numbers them, and segments them. Similarly, the device identifies each of eight maxillofacial anatomy structures in a CBCT scan, and segments them. The software facilitates navigation through the images for detailed evaluation and produces multi-planar reconstruction (MPR) views of each segmented object. The device generates a segmentation report from the input CBCT scan, for the healthcare provider's (HCP) use to further evaluate a patient's teeth and anatomy.

Here's an analysis of the acceptance criteria and study detailed in the provided FDA 510(k) clearance letter for Segmentron Viewer, organized as requested:

Acceptance Criteria and Device Performance

Device Name: Segmentron Viewer

Note: The FDA summary states that the reported Dice Coefficients exceeded the pre-defined performance goal for Tooth and Pulp Segmentation, and exceeded their respective pre-defined PGs for Anatomy Segmentation. While the specific numerical acceptance criteria (PGs) are not explicitly provided in this summary, the clearance indicates they were met.

Study Details

1. Sample Size Used for the Test Set and Data Provenance:

   • Tooth Segmentation: 126 CBCT scans (retrospective)
   • Pulp Segmentation: 43 CBCT scans (retrospective)
   • Anatomy Segmentation: 56 CBCT scans (retrospective)
   • Labeling Performance: 40 CBCT scans (from the larger validation dataset, retrospective)
   • Data Provenance: "sourced from a variety of geographic regions and demographics." It is a retrospective study.

2. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts:

   • Number of Experts: Not explicitly stated, but plural ("radiologists") is used in the summary.
   • Qualifications of Experts: U.S. board-certified radiologists. No specific years of experience are mentioned.

3. Adjudication Method for the Test Set:

   • Not explicitly stated. The summary mentions "U.S. board-certified radiologists established a reference standard for each CBCT image." This implies a single consensus or primary expert approach, but does not detail a formal adjudication process like 2+1 or 3+1.

4. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done:

   • No, a MRMC comparative effectiveness study was not reported. The studies described are "standalone validation studies," focusing on the algorithm's performance against a ground truth.

5. If a Standalone (i.e. algorithm only without human-in-the-loop performance) was done:

   • Yes, standalone validation studies were performed. The performance data section explicitly states, "DGNCT LLC evaluated the performance of Segmentron Viewer in four retrospective standalone validation studies."

6. The Type of Ground Truth Used:

   • Expert consensus, specifically manual segmentation (for segmentation studies) or annotation (for labeling studies) by U.S. board-certified radiologists.

7. The Sample Size for the Training Set:

   • Not provided in the summary. The summary describes the validation studies but does not detail the training set used for the artificial neural network models.

8. How the Ground Truth for the Training Set was Established:

   • Not provided in the summary. The summary mentions "supervised machine learning" for the algorithm, which implies a labeled training set was used, but it does not describe how these labels/ground truth were established for the training data.

Ask a specific question about this device