From: AI in Echocardiography: State-of-the-art Automated Measurement Techniques and Clinical Applications
| Company | Software package | AI-empowered tools |
|---|---|---|
| Siemens Medical Solutions Inc., USA | syngo Auto Left Heart, Acuson S2000 US System | Auto EF, Auto LV and LA volumes, Auto Strain for manually selected views |
| GE Healthcare, Inc., USA | Ultra Edition Package, Vivid Ultrasound Systems | Auto EF, Auto LV and LA volumes, Auto Strain for manually selected views |
| TOMTEC Imaging Systems GmbH, Germany | Tomtec-Arena/Tomtec-Zero | Auto EF, Auto LV and LA volumes, Auto Strain for manually selected views |
| Ultromics Ltd., United Kingdom | Echo Go/Echo Go Pro | Auto EF, Auto LV and LA volumes, Auto Strain, Auto identification of CHD (fully automated) |
| Dia Imaging Analysis Ltd., Israel | DiaCardio’s LVivoEF Software/LVivo Seamless | Auto EF and Auto Standard Echo View Identification (fully automated) |
| Caption Health, Inc., USA | The Caption Guidance Software | AI tool for assisting to capture images of a patient’s heart |
| Butterfly Network, USA | Butterfly Garden | Auto EF, Auto Standard Echo View Identification, etc. |
| US2.ai, Singapore | US2.ai | Auto Standard Echo and Strain (fully automated) |
| Report generation based on guideline criteria |
From: AI in Echocardiography: State-of-the-art Automated Measurement Techniques and Clinical Applications
| Authors | Year | Target | Measurement | Vendor | Manual/automated measurement time | Time saved (%) | Notes |
|---|---|---|---|---|---|---|---|
| Knackstedt et al.(16) | 2015 | 255 patients | EF and LS | TomTec | Manual: Not specified | - | The fully automated system provided rapid and reproducible EF and LS measurements with 0% variability in automated measurements. Good agreement with manual methods was observed. |
| Automated: 8 ± 1 s | |||||||
| Lang et al.(10) | 2021 | 200 subjects | 16 parameters LVDd, LVDs, IVS, LVPW, LVOTd, LVOT-VTI, LVEDV (A2C, A4C), LVESV (A2C, A4C), LAV (A2C, A4C), E, A, e’ (sep, lat), |
CNN model | Average | 41% | Reduced the variability of most parameters to below 10%. |
| 11′33″/6′48″ | |||||||
| Mor-Avi et al.(31) | 2023 | 12 subjects by ten experts | 20 parameters LVDd, LVDs IVS, LVPW, LVOTd, LVOT-VTI, LVEDV (A2C, A4C), LVESV (A2C, A4C), EF, LAV (A2C, A4C), E, A, e’ (sep, lat) |
Novel AI software developed collaboratively by TOMTEC | Average | 43% | DL algorithm showed good agreement with reference technique. Manual revisions improved accuracy slightly. Significant reduction in inter-reader variability. |
| 12′00″/6′49″ | |||||||
| Olaisen et al.(27) | 2024 | 50 consecutive patients | LVEDV, LVESV, and EF | Novel AI software (real-time application) | Median | 77% | Test-retest reproducibility was superior in inter-observer scenarios and non-inferior in intra-observer scenarios. AI measurements showed good agreement with reference measurements in both real-time and large research databases. |
| 7′30″/1′54″ | |||||||
| Hirata et al.(38) | 2024 | 23 consecutive patients with varying image quality and conditions by expert | 30 parameters LVDd, LVDs IVS, LVPW, LVEDV (A2C, A4C), LVESV (A2C, A4C), LAV (A2C, A4C), EF, SV, LVOTd, E, A, DT, e’ (sep, lat), a’ (sep, lat), s’(sep, lat), TRV, TAPSE, TAM, LVOT VTI, LVOT peakV, RVOT peakV, AoVmax |
US2.ai software | Average | 51% | Significant time reduction observed, especially with a good image quality. Manual adjustments required for poor image quality. |
| 5′25″/2′39″ | |||||||
| Shiokawa et al.(37) | 2024 | 30 consecutive patients | LVDd, LVDs IVS, LVPW, E, A, DT, e’ (sep, lat), a’ (lat), LVOT VTI, LVOT peakV | Philips Healthcare | Average | 27.6% | AI significantly reduced measurement time for experts and beginners, less so for intermediates. |
| 1′22″/0′59″ |
