Linking statistical shape models and simulated function in the healthy adult human heart

Area of Science:

• Computational biology
• Medical imaging
• Cardiovascular research

Background:

• Cardiac anatomy significantly influences cardiac function, but the precise impact of localized anatomical changes on functional outputs remains poorly understood.
• Statistical shape models (SSMs) offer a way to represent anatomical variations, but their direct correlation with functional simulation outcomes needs further investigation.

Purpose of the Study:

• To test the hypothesis that modes most relevant for describing cardiac anatomy in an SSM are also most important for determining electromechanical simulation outputs.
• To identify specific anatomical modes that influence different cardiac electrical and mechanical functional measures.

Main Methods:

• Creation of patient-specific four-chamber heart meshes (n=20) from cardiac CT images of asymptomatic subjects.
• Development of an SSM from 19 cases, identifying nine modes capturing 90% of anatomical variation.
• Correlation analysis and global sensitivity analysis to link SSM modes with cardiac electromechanics simulation outputs.

Main Results:

• Functional simulation outputs showed moderate correlation with modes 2, 3, and 9 (average R = 0.49, 0.37, 0.34).
• Modes 2 and 9 were most influential for left ventricular mechanics and pressure-derived phenotypes, explaining significant variance.
• Electrophysiological biomarkers were influenced by the interaction of approximately three modes, indicating complex relationships.

Conclusions:

• In healthy adult human hearts, modes explaining substantial anatomical variance do not proportionally explain electromechanical functional variation.
• Representing patient anatomy with a limited number of anatomical variation modes in cardiac models can lead to inaccuracies in simulated electromechanical function.