Automated strain measurements with AutoStrain LV/RV/LA
By Verena Roediger, PhD ∙ Jan 31, 2023 ∙ 3 min read
Global longitudinal strain (GLS) measured by echocardiography is increasingly recognized as a more effective technique than conventional ejection fraction (EF) in detecting subtle changes in left ventricular (LV) function and in predicting outcomes.1,2 It’s especially important in monitoring cancer patients undergoing chemotherapy and informing their treatment plan. Additionally, it’s critical that GLS measurement occurs in a fast and reproducible manner directly on the ultrasound system so that it can be applied to everyday clinical practice.
AI applications, such as AutoStrain, that are embedded in ultrasound systems can enable robust, reproducible one button GLS measurements. The implementation of these automation tools makes these measurements available for routine clinical practice.
At-a-glance
AutoStrain AI technology
TOMTEC, a Philips technology, has a long history of providing strain measurements with its software applications, which are well recognized and accepted by clinical users and supported by hundreds of journal publications. AutoStrain is the first TOMTEC application integrated on the Philips EPIQ CVx ultrasound system. AutoStrain, powered by Auto View Recognition, Auto Contour Placement and speckle tracking, enables a robust, reproducible one button GLS measurement, making it an efficient tool for routine clinical use.
Deformation imaging to assess left atrial (LA) and right ventricular (RV) function is a newer clinical area. AutoStrain LA and AutoStrain RV follow the Strain Task Force standardization3 document. Based on its fast and easy workflow, AutoStrain makes these measurements available for clinical routine practice.
Automation
The AutoStrain application utilizes two automation technologies: Auto View Recognition and Auto Contour Placement. While the implementation of these automation tools drives simple, fast workflow for robust and reproducible GLS measurements, the user retains the ability to edit and override the automation to ensure good clinical practice.
Auto View Recognition
Auto View Recognition automatically identifies which selected image is apical 4 chamber (A4C), apical 2 chamber (A2C) and apical 3 chamber (A3C), and automatically assigns the labels to the selected images. The label is shown on the image as a schematic overlay as shown (Figure 1). The algorithm has been validated on more than 6,000 clinical images with a success rate of 99%. This means that only 1 out of 100 cases will require manual intervention.
Auto Contour Placement
A specialized contour detection module for the respective apical view is applied to each of the three image sequences and operates in three steps. First, a complete R-R cycle – end-diastole (ED) start to ED end – is selected from each sequence. Second, in the ED start frame of that cycle, the left ventricle (LV) is automatically located. Third, a view-specific deformable endocardial contour model is aligned to the individual image content (Figure 2). The same approach is applied for RV Auto Contour Placement while the initial LA contour is placed in the end-systolic (ES) frame (Figure 2).
Gallery
- Toggle view
Figure 2. LV focused view with Auto Contour Placement, LA focused view with Auto Contour Placement, RV focused view with Auto Contour Placement.
Speckle tracking
Once the endocardial border is automatically placed in ED (LV, RV) or in ES (LA), it follows the cardiac motion using speckle tracking over the entire cardiac cycle. In Tracking Revision and Analysis Workflow step, the user has all the measurements displayed along with the ED and ES borders next to a dynamic display of all views to easily verify that the borders are correctly placed and tracked. If border editing is needed, it is highly recommended to start editing in ED. Editing the border in ED triggers new speckle tracking of the border throughout the cardiac cycle. When editing the ES border, the edits are propagated to the neighboring frames but the ED border remains untouched.
Strain measurement
Left ventricle (LV)
Longitudinal strain is measured at the endocardial border as indicated by the green line. Instantaneous endocardial strain is visualized by color-coding close to the endocardial border.
The ED frame is always the first frame in the selected cardiac cycle. The ES time point is automatically defined as the time point of global peak strain. It can be adjusted according to aortic valve closure (AVC) time in the AVC layout.
GLS is calculated as global shortening of the endocardial border. It is defined as a peak value and thus independent from AVC as shown (Figure 3).
Right ventricle (RV)
Based on the deformation of the green endocardial contour, longitudinal strain is computed for the free wall (RVFWSL), the global 4-chamber contour (RV4CSL) and the three free wall segments.
Left atrial (LA)
The LA cardiac cycle consists of three phases: reservoir phase goes from ED to ES, conduit phase ends at the time point right before atrial contraction (AC) – also denoted as PreA time point – and contraction phase completes the cardiac cycle.
Gallery
- Toggle view
Figure 3. Segmental strain curves and segmental strain values displayed at peak-systolic and end-systolic strain. GLS at peak strain is visualized with the global strain curve.
Validation
Validation of AutoStrain LV has been done in comparison to the well-accepted TOMTEC 2D CPA application, with 225 clips analyzed with 2D CPA and re-evaluated with AutoStrain LV using consistent border definitions and manual corrections where necessary in both packages.
For AutoStrain LA validation, LA reservoir strain with reference frame at ED (LASr_ED) has been compared to 2D CPA using 71 clips. AutoStrain RV compared RV global 4 chamber longitudinal strain (RV4CSL) in 75 clips with the corresponding 2D CPA measurements.
As deformation within the myocardium varies regionally, GLS as a measure of myocardial deformation shows interdependence with initial contour placement. To assess reproducibility of the AutoStrain tracking, the influence of manual adjustments of the initial endocardial contour was evaluated. Based on 225 user-reviewed initial contours, manual refinements have been simulated by slight modification of all original contour point positions (within ±2 pixels distance). Then resulting GLS measurements were compared against the original readings.
Verena Roediger, PhD
Product Manager, TOMTEC
Munich, Germany
Clinical article
AutoStrain LV/RV/LA – automated strain measurements
Subscribe to our email updates
Footnotes
1 Voigt JU, et al., Definitions for a common standard for 2D speckle tracking echocardiography: consensus document of the EACVI/ASE/Industry Task Force to standardize deformation imaging. J Am Soc Echocardiogr. 2015;28(2):183-93. doi: 10.1016/j.echo.2014.11.003. 2 Biering-Sorensen T, Biering-Sorensen S, Olsen FJ, et al. Global Longitudinal Strain by Echocardiography Predicts Long-Term Risk of Cardiovascular Morbidity and Mortality in a Low-Risk General Population The Copenhagen City Heart Study. Cardiovasc Imaging. 2017:10. doi: 10.1161/CIRCIMAGING.116.005521. 3 Badano LP, et al. Standardization of left atrial, right ventricular, and right atrial deformation imaging using two-dimensional speckle tracking echocardiography: a consensus document of the EACVI/ASE/Industry Task Force to standardize deformation imaging. Eur Heart J Cardiovasc Imaging. 2018 Jun;19(6):591-600. doi: 10.1093/ehjci/jey042.
Disclaimers
Results are specific to the institution where they are obtained and may not reflect the results achievable at other institutions.
