Search Results

The X9 Ultrasound System is intended for ultrasound imaging of validated vascular accesses for hemodialysis during the preliminary stage of a cannulation procedure, prior to inserting the needle. The device is intended to be used by qualified healthcare professionals in a medical setting. The device is not intended to be used to diagnose the position of the vascular access without confirming position per standard of care. The X9 Ultrasound System is not for use in pediatric patients.

The X9 Ultrasound System is intended for guidance in accessing arteriovenous fistulas and grafts (AVF/G). The X9 Ultrasound System consists of a Handpiece and Software. The Handpiece is a handheld device with an ultrasound transducer and is attached via a cable to a user-supplied Computer. The Handpiece is covered by a compatible probe cover and is placed on the patient's limb and positioned (translated/rotated) by the user to align with the AVF/G. The Handpiece contains a physical button for the user to activate/deactivate the system and an alignment marking to assist the user with positioning the Handpiece over the vessel. The software, which runs on a standard operating system platform installed on the Computer, provides the Graphical User Interface (GUI). The GUI will indicate when the Handpiece is aligned with the AVF/G. The information given from the system is intended to guide the user so that they can efficiently proceed with the standard of care assessment prior to cannulation.

Here's a detailed breakdown of the acceptance criteria and the study that proves the X9 Ultrasound System meets those criteria, based on the provided FDA 510(k) clearance letter:

Acceptance Criteria and Device Performance

1. Table of Acceptance Criteria and Reported Device Performance

Note: The FDA 510(k) summary explicitly states that the Machine Learning Model (and therefore the X9 Ultrasound System) met the ML Model Validation acceptance criteria for the primary analyses.

Study Details

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

• Sample Size for Test Set: 378 B-mode ultrasound images collected from 63 participants.
• Data Provenance: The study involved participants known to have an arteriovenous fistula or graft (AVF/G). Clinic locations for ML model validation were selected to obtain a sample of participants. The patient demographics provided (31 females, 32 males, 34 Black/African American, 27 White, 1 Asian, 1 Unknown/Not Reported, 4 Hispanic or Latino, and 59 Not Hispanic or Latino, aged 18 to 75+ years old) suggest a diverse, likely multi-center, prospective collection of data within a clinical setting, although specific countries are not mentioned. Given the FDA submission, the data is most likely from the United States or a country adhering to similar clinical trial standards.

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

• Number of Experts: Three (3) independent sonographers.
• Qualifications of Experts: The document states they used their "clinical expertise" to determine if an access vessel was visible and to annotate its location. This implies they are trained and experienced medical professionals in ultrasound imaging, specifically sonographers.

4. Adjudication Method for the Test Set

• The ground truth for Access Presence was established when the "majority of independent reviewers determined that an access vessel was present in the image." This indicates a 2 out of 3 consensus (or "2+1") method for access presence.
• For the True Location of the access vessel, it was established by calculating the average of the locations annotated by each of the independent reviewers. This is an averaging adjudication method for continuous metrics.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

• No, a MRMC comparative effectiveness study was not explicitly done or reported. The study focused on the standalone performance of the machine learning model against expert-annotated ground truth. There is no mention of comparing human readers with AI assistance versus human readers without AI assistance, nor any effect sizes of human reader improvement.

6. Standalone (Algorithm Only) Performance

• Yes, a standalone performance study was done. The "Machine Learning Model Validation" directly assesses the performance of the algorithm (X9 Ultrasound System's ML model) in determining access vessel presence and location against expert-established ground truth. The reported metrics (Access Presence Sensitivity, Average Lateral Error, and Average Dice Similarity Coefficient) are all measures of the algorithm's performance alone.

7. Type of Ground Truth Used

• The ground truth used was expert consensus and expert annotation.
  • For "Access Presence," it was based on the majority decision of three independent sonographers.
  • For the "True Location" of the access vessel, it was the average of the locations annotated by the three independent sonographers.

8. Sample Size for the Training Set

• The document states, "The training dataset was collected at dialysis clinics not included in the ML model validation and was used exclusively for model training." However, the specific sample size (number of images or participants) for the training set is not provided in this document.

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

• The document states, "Separate test datasets were reserved for performance validation against expert-annotated ground truth." This implies that the training data also likely utilized expert-annotated ground truth, similar to the test set, but this is not explicitly detailed for the training set. It's standard practice for machine learning models to be trained on data with established ground truth, often through expert labeling or external verified sources.

Ask a specific question about this device