Search Results

UltraExtend NX CUW-U001S Ultrasound Image Analysis Program is designed to allow the user to observe images and perform analysis based on examination data acquired using the following diagnostic ultrasound systems; TUS-AI900, TUS-AI800, and TUS-AI700.

This system is suitable for use in hospital and clinical settings by physicians or legally qualified persons who have received the appropriate training.

Device Description

The UltraExtend NX, V2.0 is designed to allow the user to observe images and perform analysis based on examination data acquired using the Aplio i900/i800/i700 diagnostic ultrasound systems. RAW only or data saved in Image + RAW should be used for UltraExtend NX.

AI/ML Overview

The FDA 510(k) clearance letter for the UltraExtend NX CUW-U001S V2.0 Ultrasound Image Analysis Program indicates that the device has integrated AI/ML-based functionality (2D Wall Motion Tracking with Full-assist function for left ventricle (LV) and Auto EF with Full-assist function for LV) that was previously cleared with a reference device (K223017). The submission states that "these studies utilized a representative subset of the clinical data acquired for the original performance testing of these features; additionally these studies applied the same acceptance criteria to evaluate the performance of these features compared to the same ground truth as utilized in the original performance evaluation of these features with the reference device."

Unfortunately, the provided text does not contain the specific acceptance criteria or detailed results of the performance testing for these AI/ML features. It only states that the features "perform as intended when integrated into the subject device, and with substantial equivalence as with the reference device."

Therefore, I cannot provide a table of acceptance criteria and reported device performance or many of the specific details requested in your prompt based solely on the provided document. The document refers to the original performance testing of the reference device (K223017) for these details.

However, I can extract and infer information about the study design to the extent possible:

Here's what can be inferred from the provided text, and what cannot be determined:

Acceptance Criteria and Device Performance

The document states that the same acceptance criteria as the original performance testing for the reference device (K223017) were applied.
Cannot Determine: The specific numerical acceptance criteria (e.g., specific accuracy, sensitivity, specificity thresholds) or the reported device performance metrics (e.g., actual accuracy, sensitivity, specificity values) are not provided in this document.

Study Information

Information Type	Details from Document
1. Acceptance Criteria & Reported Performance	Acceptance Criteria: "applied the same acceptance criteria to evaluate the performance of these features compared to the same ground truth as utilized in the original performance evaluation of these features with the reference device."
Reported Performance: "The results of this testing demonstrate that both features perform as intended when integrated into the subject device, and with substantial equivalence as with the reference device."
No specific numerical criteria or performance values are provided.
2. Sample Size (Test Set) & Data Provenance	Sample Size: "a representative subset of the clinical data acquired for the original performance testing of these features"
The exact number of cases/samples in this subset is not specified.
Data Provenance: "clinical data"
Country of origin (likely global, given the company's international presence but not explicitly stated), and whether retrospective or prospective is not explicitly stated for the test set, but "acquired" suggests previously collected.
3. Number & Qualifications of Experts	Cannot determine. The document does not specify the number or qualifications of experts used for establishing the ground truth or for any readouts.
4. Adjudication Method (Test Set)	Cannot determine. The method used for adjudicating expert opinions to establish ground truth (e.g., 2+1, 3+1) is not provided.
5. MRMC Comparative Effectiveness Study	Not an MRMC Study. The testing described is not a multi-reader multi-case comparative effectiveness study comparing human readers with and without AI assistance. It is focused on demonstrating the embedded AI/ML features perform as intended and substantially equivalent to their performance in the previous device. There's no mention of human reader efficacy improvement.
6. Standalone Performance (Algorithm Only)	Yes, indirectly. The performance evaluation of the AI/ML-based functionality (2D Wall Motion Tracking with Full-assist function for left ventricle and Auto EF with Full-assist function for left ventricle) within the UltraExtend NX device is focused on how the integrated features perform, compared to the ground truth. While it's integrated into a user-facing product, the "Full-assist function" implies an algorithmic component being evaluated against a ground truth. The submission confirms "the results of this testing demonstrate that both features perform as intended when integrated into the subject device".
7. Type of Ground Truth Used	"the same ground truth as utilized in the original performance evaluation of these features with the reference device." No further specifics on the nature of the ground truth (e.g., expert consensus, pathology, follow-up outcomes) are provided.
8. Sample Size (Training Set)	Cannot determine. The document does not provide any information about the training set size for the AI/ML models. It only discusses the test set used for the validation of the integrated features.
9. How Ground Truth for Training Set Established	Cannot determine. Given that the training set details are not provided, how its ground truth was established is also not present in this document.

Summary of missing information:

To fully answer your prompt, you would need to consult the original 510(k) submission for the reference device (K223017), Aplio i900/i800/i700 Diagnostic Ultrasound System, Software Version 7.0, as that is where the detailed performance data, acceptance criteria, and ground truth establishment methodology for the AI/ML features would have been submitted and evaluated by the FDA. The current document (K250328) focuses on demonstrating that these already cleared AI/ML features maintain their performance when integrated into a new workstation.

Ask a Question

Ask a specific question about this device

Page 1 of 1