Study List Filters
A study list filter is available to search for studies with a specific icardio.ai algorithm status. This filter is called ‘AI Status’.
The status of AI on an exam can also be seen in the study information section under ‘AI Status’.
As AI is initiated, either manually or automatically, the exam will progress through a variety of Statuses.
| Initiating | Initiated in UltraLinQ |
| Pending | Contact made with icardio.ai; Wait for results to return. |
| Completed | Inference has been completed; Results are available in Viewer and AI Report tab. |
| Failed | Inference has failed. Contact UltraLinQ Support team for assistance. |
| Ineligible | Exam does not meet all technical requirements for icardio.ai algorithms. Contact your Client Success representative for assistance. |
Viewing icardio.ai Results in UltraLinQ
Quantitative AI
Results of the Quantitative AI algorithm, which presents automated measurements of the echo, are displayed in the following manner.
| Thumbnail Indicator |
|
| Results Table |
|
| AI Report |
|
Echo Check AI
| Thumbnail Indicator |
|
| Viewer Score |
|
| AI Report |
|
Quality Score
Quality Scores are currently only run on the A2C view, the A4C view, and PLAX views.
The full criteria when analyzing and producing Study Quality are as follows:
- Zoom level
- Gain/Clarity
- Angulation
- Foreshortening
- Presence or absence of anatomical structures
- Endocardial clarity and definition
- Potential for image quality to jeopardize accuracy or reliability of linear measurements or chamber tracings.
- Potential for image quality to mislead clinical assumptions, for instance clarity of valvular leaflets.
- The cine loop contains a full heartbeat, displaying each phase of the cardiac cycle.
Quantitative AI uses views that are considered to be the most reliable when generating certain measurements. If these measurements are not available in the Results, it is likely that these views are missing or are of suboptimal quality.
Perspective Confidence Score
Confidence Score is EchoCheck's confidence in determining a viewport by the image provided and analyzed. In echos, certain views can indeed be easily confused with one another, especially when the image acquisition is suboptimal.
Perspective Confidence is a score from 0% to 100% that represents how confident the AI model is in its view classification. The score is determined by the B-mode classifier model's output. The code uses threshold values to determine if a perspective is acceptable--anything below 30% confidence score is rejected, and no additional pipelines will be run.
An 80% score for Confidence Score for PLAX means that EchoCheck is 80% confident that the view analyzed is PLAX and is 20% confident that it is a different view, like PLAX Aortic Cusps or PLAX Mitral Cusps. Other examples:
- A4C (Apical 4-Chamber): This view is distinct because it shows all four chambers of the heart, but it can be confused with the A5C (Apical 5-Chamber) view if the acquisition includes the outflow tract, thus adding a fifth 'chamber' to the view.
- A3C (Apical 3-Chamber) and A2C (Apical 2-Chamber): These views are obtained by rotating the transducer from the apical four-chamber position and can be confused with each other if the rotation or angulation is not precise.
Results Table
The Results Table will contain all results from the Echo Check and Quantitative components of the icardio.ai algorithm. This table is accessible through the Table icon in the top-right corner of Viewer.
The Results Table will open in the top-left of the Viewer.
The Results are ‘actively linked’ to the images of the exam. This means a user can open the image that score or value from icardio.ai is associated with if it is clicked in the Results Table.
Hint: Using the split-screen view in UltraLinQ's viewer, users can visualize the table and their Worksheet simultaneous.
AI Report tab
A full ‘AI Report’ of the icardio.ai run on a particular study can be found in its ‘AI Report’ tab.
A full AI Report contains:
- Results from the Echo Check and Quantitative AI algorithms
- Conclusions of the algorithms
- Reference ranges for the measurements, based on the Gender of the exam's patient
- Quantitative AI Overlays of the image measurements.
Example:
| DICOM ID | Unique ID of the exam |
| View Type |
The viewport identified by the Echo Check algorithm. Problematic view types like "Unclear Dark," "Unclear Noisy," and "Other," are automatically flagged as unsuitable for measurements. |
| View Confidence |
View Score represents the confidence of the algorithm in the View Type. A color-coding system is included to quickly identify issues or areas of concern for the exam.
|
| Quality Score |
Quality Score represents an objective score of the image. A color-coding system is included to quickly identify issues or areas of concern for the exam.
|
| Notes | Notes provide explanations when measurements may be unavailable or require caution. |
| Footnotes | Reference information about thresholds, special cases, and why quality scores are not available. |