Search Results

The product is designed to provide fluoroscopic and digital spot images of adult and pediatric patient populations during diagnostic, interventional, and surgical procedures. Examples of a clinical application may include: orthopedic, gastrointestinal, endoscopic, urologic, neurologic, vascular, critical care, and emergency procedures.

The OEC One CFD is a mobile C-arm X-ray system to provide fluoroscopic images of the patient during diagnostic, interventional, and surgical procedures such as orthopedic, gastrointestinal, endoscopic, urologic, neurologic, vascular, critical care, and emergency procedures. These images help the physician visualize the patient's anatomy and localize clinical regions of interest. The system consists of a mobile stand with an articulating arm attached to it to support an image display monitor (widescreen monitor) and a TechView tablet, and a "C" shaped apparatus that has a flat panel detector on the top of the C-arm and the X-ray Source assembly at the opposite end.

The OEC One CFD is capable of performing linear motions (vertical, horizontal) and rotational motions (orbital, lateral, wig-wag) that allows the user to position the X-ray image chain at various angles and distances with respect to the patient anatomy to be imaged. The C- arm is mechanically balanced allowing for ease of movement and capable of being "locked" in place using a manually activated lock.

The subject device is labelled as OEC One CFD.

Here's a summary of the acceptance criteria and the study details for the OEC One CFD device, specifically focusing on the new "Deep learning-based Trajectory Pointer software" feature, as that's where the most detailed study information is provided in the document.

1. Table of Acceptance Criteria and Reported Device Performance (for Trajectory Pointer)

2. Sample Size Used for the Test Set and Data Provenance

• Total image dataset: 3078 images
• Test set size: 307 images (approx. 10% of the total dataset)
• Data Provenance (Ethnicity/Country):
  • US: 51.1%
  • China: 35.8%
  • Phantom: 9.8%
  • EU/Australia: 3.3%
• Data type: The document states that sample clinical images were evaluated to demonstrate substantial equivalence, and "all images were collected from GE OEC Mobile C-arm devices," suggesting retrospective collection of clinical images, alongside phantom data. It also notes that the test data covers the "entire range of values for all confounders," implying it's representative of real-world scenarios.

3. Number of Experts Used to Establish Ground Truth for the Test Set and Qualifications

• Number of Experts: 3
• Qualifications: U.S. board-certified radiologists. (No specific years of experience are listed).

4. Adjudication Method for the Test Set

• The document states: "3 U.S. board certified radiologists check annotations to make sure the ground truth is correct."
• This implies a consensus or verification process by multiple experts to validate the initial manual annotations, but a specific method like "2+1" or "3+1" is not explicitly mentioned.

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

• The provided text does not indicate that a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was done to evaluate human reader improvement with AI assistance. The performance metrics reported are for the algorithm's standalone performance.

6. If a Standalone (Algorithm Only) Performance Study was Done

• Yes, a standalone performance study was done for the Trajectory Pointer feature. The reported success rates of 98.8% for LIH images and 97.1% for Fluoroscopy images (both exceeding the 95% MAAP) are measures of the algorithm's performance.

7. The Type of Ground Truth Used

• For the Trajectory Pointer feature, the ground truth was established by expert consensus/manual annotation. Specifically, "the contours of K-wires were manually outlined by trained annotators," and "3 U.S. board certified radiologists check annotations to make sure the ground truth is correct."

8. The Sample Size for the Training Set

• Training set size: 2771 images (3078 total images - 307 test images = 2771 images). The document states: "Total dataset included 3078 images, 307 were used for test dataset and the rest for training/tuning."

9. How the Ground Truth for the Training Set was Established

• While not explicitly detailed for the training set, it is highly probable that the ground truth for the training data was established in a similar manner to the test set: through manual outlining of K-wire contours by trained annotators. The document mentions "comparing ground truth annotations" during deduplication, implying annotations were available for the training data.

Ask a specific question about this device