Left ventricular modelling: a quantitative functional assessment tool based on cardiac magnetic resonance imaging

Area of Science:

• Cardiology
• Medical Imaging
• Biomedical Engineering

Background:

• Left ventricle (LV) dysfunction diagnosis is crucial for cardiovascular health.
• Cardiac magnetic resonance (CMR) provides detailed LV imaging.
• Manual analysis of CMR data can be time-consuming and subjective.

Purpose of the Study:

• To develop and validate a semi-automated tool for diagnosing LV dysfunction from CMR.
• To compare the tool's performance against manual analysis by an expert cardiologist.
• To assess the tool's capability in quantifying LV function, including strains and torsion.

Main Methods:

• Development of a semi-automated tool processing CMR short-axis images.
• Detection of endocardial and epicardial contours for volume and mass calculation.
• Integration of nearest neighbour tracking and finite-element theory for local myocardial strain and torsion analysis.

Main Results:

• Feasible automated analysis in 13/15 patients; good correlation of volumes and mass with manual data.
• High sensitivity (77.8%), specificity (85.1%), and accuracy (82%) in detecting regional wall motion abnormalities.
• Accurate strain calculation on a virtual phantom and demonstration of reduced torsion in pathological LVs.

Conclusions:

• The semi-automated tool is a valid approach for fast LV function analysis from CMR.
• The method shows potential for improved diagnostic efficiency in cardiology.
• Further improvements are possible to enhance the tool's capabilities.