Comparison of manual and artificial intelligence based quantification of myocardial strain by feature tracking—a cardiovascular MR study in health and disease - Scorecard - MDSpire
Advertisement
Comparison of manual and artificial intelligence based quantification of myocardial strain by feature tracking—a cardiovascular MR study in health and disease
Clinical Scorecard: Evaluation of Manual Versus AI-Driven Myocardial Strain Measurement Through Feature Tracking: A Cardiovascular MRI Investigation in Healthy and Diseased States
At a Glance
Category
Detail
Condition
Myocardial strain assessment in healthy and diseased cardiac states
Key Mechanisms
Quantitative myocardial deformation measurement via cardiovascular magnetic resonance (CMR) feature tracking (FT) using manual or AI-driven segmentation
Target Population
Healthy volunteers and patients with various cardiac diseases including LV hypertrophy, aortic stenosis, hypertrophic cardiomyopathy, and chronic myocardial infarction
Care Setting
Cardiovascular imaging and clinical routine using CMR in hospital or specialized cardiac centers
Key Highlights
Feature tracking (FT) on standard cine CMR images enables assessment of longitudinal, circumferential, and radial myocardial strain without special sequences.
Manual contouring for FT is time-consuming and influenced by reader expertise; AI-driven segmentation offers potential for standardization and efficiency.
AI segmentation models trained on diverse datasets including pathological cases improve reproducibility and may facilitate consensus on FT methodology.
Guideline-Based Recommendations
Diagnosis
Use CMR feature tracking to quantify global and segmental myocardial strain in longitudinal, circumferential, and radial directions.
Apply the 17-segment American Heart Association model (excluding apical segment) for strain analysis.
Exclude slices with visible left ventricular outflow tract (LVOT) from strain analysis.
Management
Consider AI-based segmentation to reduce manual contouring time and variability in strain assessment.
Validate AI-generated contours manually to ensure accuracy before strain calculation.
Use steady-state free precession cine sequences at 1.5T or 3T scanners for image acquisition.
Monitoring & Follow-up
Assess strain values longitudinally to detect myocardial changes even in preserved left ventricular function.
Monitor tracking quality by evaluating mesh overlay and myocardial point tracking through cardiac phases.
Risks
Manual contouring variability may affect strain measurement accuracy.
Inclusion of LVOT in contours can lead to inaccurate strain values.
Respiratory artifacts and incomplete ventricular coverage can compromise data quality.
Patient & Prescribing Data
Healthy adults and patients with left ventricular hypertrophy, aortic stenosis, hypertrophic cardiomyopathy, and chronic myocardial infarction
AI-driven FT strain assessment may enhance diagnostic accuracy and efficiency in clinical routine and research by reducing manual workload and standardizing measurements.
Clinical Best Practices
Perform endo- and epicardial contouring at end-diastole phase for both manual and AI segmentation.
Avoid contouring phases with visible LVOT to prevent strain measurement errors.
Manually validate AI-generated reference points and contours prior to strain calculation.
Use consistent software and protocols to minimize variability in strain values.
Apply FT algorithms that track myocardial points throughout the cardiac cycle to ensure accurate strain derivation.
