Analysis of nonstandardized stress echocardiography sequences using multiview dimensionality reduction

Area of Science:

• Cardiology
• Medical Imaging
• Data Science

Background:

• Current stress echocardiography protocols face limitations in large-scale screening.
• Handgrip exercise offers a promising alternative but presents challenges in data analysis due to complex, large datasets.

Purpose of the Study:

• To develop and validate a computational framework for analyzing complex handgrip stress echocardiography data.
• To explore the potential of this framework in differentiating patient groups and understanding disease mechanisms.

Main Methods:

• An unsupervised multiple kernel learning approach was employed to integrate myocardial velocity and heart rate data.
• Multiscale kernel regression was used for reconstructing population-specific signatures.
• Clustering algorithms analyzed projected data sequences to identify subject groups.

Main Results:

• The framework successfully analyzed a dataset of 10 healthy controls and 5 ANT1 mutation patients.
• Distinctive clusters of response were identified, differentiating between the groups.
• The analysis provided insights into the underlying pathophysiology of ANT1 mutation.

Conclusions:

• The proposed framework offers a robust method for analyzing complex stress echocardiography datasets.
• This approach facilitates large-scale screening and enhances understanding of cardiovascular conditions.
• The findings support the clinical utility of handgrip stress echocardiography with advanced data analysis.