(265 days)
Sonix Health is intended for quantifying and reporting echocardiography for use by or on the order of a licensed physician. Sonix Health accepts DICOM-compliant medical images acquired from ultrasound imaging devices. Sonix Health is indicated for use in adult populations.
Sonix Health comes with the following functions:
- Checking ultrasound multiframe DICOM
- Echocardiography multiframe DICOM classification and automatic measurement.
- Verification of the results and making adjustments manually.
- Providing the report for analysis
Sonix Health will be offered as SW only, to be installed directly on customer PC hardware. Sonix Health is DICOM compliant and is used within a local network.
Sonix Health utilizes a two-step algorithm. A single identification model identifies a view in the first step. The second step performs the deep learning according to the view. The deep learning algorithms for the second step are categorized as B-mode, and Doppler algorithms. The main algorithm of Sonix Health is to identify the view and segment the anatomy in the image.
The provided text describes the performance evaluation of a medical device named "Sonix Health" for quantifying and reporting echocardiography. Here's a breakdown of the requested information:
Device: Sonix Health (K240645)
Software Functions:
- Checking ultrasound multiframe DICOM
- Echocardiography multiframe DICOM classification and automatic measurement.
- Verification of the results and making adjustments manually.
- Providing the report for analysis.
- Utilizes a two-step algorithm: single identification model for view recognition, followed by deep learning for B-mode and Doppler algorithms. Main algorithm identifies view and segments anatomy.
1. Table of Acceptance Criteria and Reported Device Performance
| Feature | Acceptance Criteria | Reported Device Performance |
|---|---|---|
| View Recognition | Average accuracy ≥ 84% | 96.25% average accuracy for additional views. |
| Auto Measure | Average correlation coefficient ≥ 0.80 (compared to manual measurements) | 0.918 average correlation coefficient (compared to manual measurements). |
| Auto Strain | ||
| LVGLS, LARS, LACts | Average correlation coefficient ≥ 0.80 (compared to manual measurements) | 0.88 average correlation coefficient. |
| RV Free Wall Strain | Average correlation coefficient ≥ 0.60 (compared to manual measurements) | 0.69 correlation coefficient. |
| Average GLS | RMSE ≤ 3.00% (compared to manual measurements) | 2.16% RMSE. |
| Segmental Longitudinal Strain | RMSE ≤ 7.50% (compared to manual measurements) | 6.32% RMSE. |
2. Sample Size and Data Provenance
- Total Patients: 335
- Data Provenance:
- 303 patients (90%) originated from the U.S. (Mayo Clinic in Arizona) and South Korea (Severance Hospital, Seoul).
- Specifically, 30% (93 patients) of these 303 were from U.S. hospitals.
- 70% (200 patients) of these 303 were from Korean hospitals.
- An additional 32 patients (10%) were obtained from South Korea (Severance Hospital, Seoul).
- 303 patients (90%) originated from the U.S. (Mayo Clinic in Arizona) and South Korea (Severance Hospital, Seoul).
- Recruitment Type: Images were "taken for diagnostic purposes in actual clinical settings" and "acquired following the IRB procedures," suggesting a retrospective collection of existing patient data.
3. Number and Qualifications of Experts for Ground Truth
- Experts for Annotation: Two experienced sonographers with Registered Diagnostic Cardiac Sonographer (RDCS) certification.
- Supervising Experts: Two experienced cardiologists.
4. Adjudication Method for the Test Set
- The text states, "The annotation was supervised by two experienced cardiologists and the consensus annotation was used as the final ground truth." This implies a form of consensus-based adjudication, but the exact process (e.g., if initial annotations were independent, how disagreements were resolved, etc.) is not detailed beyond "consensus annotation."
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
- The document does not describe a multi-reader multi-case (MRMC) comparative effectiveness study to evaluate "human readers improve with AI vs without AI assistance." The study focuses on evaluating the standalone performance of the AI model against expert manual measurements, and the device is intended for human-in-the-loop use where users review and modify results.
6. Standalone (Algorithm Only) Performance
- Yes, a standalone performance evaluation was primarily done. The metrics presented (accuracy, correlation coefficients, RMSE) directly assess the algorithm's output compared to ground truth, which was established by experts' manual measurements or reference devices. Although the device is designed for human review, the reported performance metrics quantify the automated capabilities of the software.
7. Type of Ground Truth Used
- The ground truth for the test set was established through expert consensus annotation.
- For strain measurements, the ground truth was "established by the experts with the help of the reference devices (EchoPAC for global longitudinal, segmental and RV free wall strain and TOMTEC Arena for LA reservoir and contraction strain)." This means the ground truth combines expert interpretation with measurements derived from established medical software.
8. Sample Size for the Training Set
- The document states, "The training data and validation data are distinct and independent." However, the sample size for the training set is not provided in the given text.
9. How Ground Truth for the Training Set Was Established
- The document explicitly states how the ground truth for the test set was established (expert consensus, aided by reference devices).
- However, the text does not describe how the ground truth for the training set was established.
§ 892.2050 Medical image management and processing system.
(a)
Identification. A medical image management and processing system is a device that provides one or more capabilities relating to the review and digital processing of medical images for the purposes of interpretation by a trained practitioner of disease detection, diagnosis, or patient management. The software components may provide advanced or complex image processing functions for image manipulation, enhancement, or quantification that are intended for use in the interpretation and analysis of medical images. Advanced image manipulation functions may include image segmentation, multimodality image registration, or 3D visualization. Complex quantitative functions may include semi-automated measurements or time-series measurements.(b)
Classification. Class II (special controls; voluntary standards—Digital Imaging and Communications in Medicine (DICOM) Std., Joint Photographic Experts Group (JPEG) Std., Society of Motion Picture and Television Engineers (SMPTE) Test Pattern).