(90 days)
Automatic Registration is a software device for image guided surgery intended to be used in combination with compatible Brainlab navigation systems such as the Brainlab Spine & Trauma Navigation System. Automatic Registration provides an image registration for intraoperatively acquired 3D CT/CBCT or fluoroscopic images.
The Subject Device Automatic Registration is an accessory to the Brainlab Spine & Trauma Navigation System. It correlates intraoperatively acquired patient data (3D CT/CBCT or fluoroscopic images) to the surgical environment in order to provide a patient registration for subsequent use by the Brainlab Spine & Trauma Navigation. The device includes the following software modules:
- Automatic Registration 2.6
- Universal Atlas Performer 6.0
- Universal Atlas Transfer Performer 6.0 .
And uses as well several hardware devices, mainly registration matrices, adhesive flat markers and a calibration phantom, for performing the registration. The software is installed on an Image Guided Surgery (IGS) platform. The registration matrices are reusable devices, delivered nonsterile and having patient contact.
Here's a breakdown of the acceptance criteria and the study details for the Automatic Registration device, based on the provided FDA 510(k) summary:
1. Table of Acceptance Criteria and Reported Device Performance
| Performance Metric | Acceptance Criteria | Reported Device Performance |
|---|---|---|
| Accuracy (New Registration Method) | Registration accuracy of ≤ 2.5 mm (P95) with a mean navigation accuracy with 3D deviation ≤ 1.5 mm | Registration accuracy of ≤ 2.5 mm (P95) with a mean navigation accuracy with 3D deviation ≤ 1.5 mm |
| Software Verification | Successful implementation of product specifications, incremental testing, risk control, compatibility, cybersecurity | Successfully conducted as recommended by FDA guidance. |
| AI/ML Landmark Detection | Quantifying object detection, quality of vertebra level assignment, quality of landmark predictions, performance of observer view direction. | Assessed by quantifying the above aspects for AI/ML detected landmarks on X-rays. |
| Usability | No critical use-related problems identified. | No critical use-related problems identified after summative usability testing. |
2. Sample Size Used for the Test Set and Data Provenance
- Accuracy Testing (New Registration Method): The testing was performed on human cadavers. The exact number of cadavers or cases within them is not specified.
- AI/ML Assessment: The summary states the algorithm was developed using a "controlled internal process that defines activities from the inspection of input data to the training and verification of the algorithm." No specific sample size for the test set is provided, nor is the data provenance (e.g., country of origin, retrospective/prospective) explicitly mentioned for the AI/ML assessment tests.
- Usability Testing: 15 representative users were used for summative usability testing.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts
- For the accuracy testing on human cadavers, the method for establishing ground truth and the number/qualifications of experts are not explicitly stated. It can be inferred that ground truth for navigation accuracy would involve precise measurements by trained personnel, likely using specialized equipment, but no details are provided.
- For the AI/ML algorithm assessment, the criteria for "quantifying object detection, quality of vertebra level assignment, quality of landmark predictions, and the performance of the observer view direction" would imply expert review. However, the number of experts and their qualifications for establishing this ground truth are not specified.
4. Adjudication Method for the Test Set
The document does not specify any adjudication methods (e.g., 2+1, 3+1) for the test sets described.
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done
No, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was not done or at least not reported in this summary. The summary focuses on the standalone performance of the device and its components, particularly the new AI/ML registration update method, against predefined accuracy criteria and usability.
6. If a Standalone (Algorithm Only Without Human-in-the-Loop Performance) Was Done
Yes, a standalone performance evaluation was done for the new registration method, specifically the AI/ML algorithm component. The "Machine Learning" section describes the assessment of the AI/ML detected landmarks on X-rays, evaluating aspects like object detection, vertebra level assignment, and landmark predictions. The accuracy bench testing also evaluates the "overall system registration accuracy" of the new method, implying its standalone performance in achieving the specified accuracy.
7. The Type of Ground Truth Used
- For accuracy testing (bench test on cadavers): The ground truth would likely be established through highly precise physical measurements on the cadavers, probably using a gold-standard measurement system (e.g., CMM, anatomical landmarks verified with high-precision tools). The document doesn't explicitly state the methodology, but this is typical for navigation accuracy.
- For AI/ML assessment: The ground truth for "object detection, quality of vertebra level assignment, quality of landmark predictions, and the performance of the observer view direction for 2D X-rays" would have been established by human experts, likely radiologists or orthopedic surgeons with expertise in spinal anatomy and imaging. The document doesn't explicitly detail the methodology or the experts involved.
8. The Sample Size for the Training Set
The sample size for the training set of the Convolutional Neural Network (CNN) is not specified in the provided document. It only mentions that the algorithm was "developed using a controlled internal process that defines activities from the inspection of input data to the training and verification of the algorithm."
9. How the Ground Truth for the Training Set Was Established
The document states that the AI/ML algorithm was developed using a "Supervised Learning approach." This implies that the training data was labeled by human experts. However, the specific method (e.g., single expert, consensus, specific qualifications of experts) for establishing this ground truth for the training set is not detailed in the provided text.
§ 882.4560 Stereotaxic instrument.
(a)
Identification. A stereotaxic instrument is a device consisting of a rigid frame with a calibrated guide mechanism for precisely positioning probes or other devices within a patient's brain, spinal cord, or other part of the nervous system.(b)
Classification. Class II (performance standards).