Direct estimation of global longitudinal strain from echocardiograms using a logarithm-scaled Fourier magnitude correlation

TL;DR

This paper introduces a novel logarithm-scaled Fourier magnitude correlation method for estimating global longitudinal strain from echocardiograms, which outperforms traditional techniques by not requiring segmentation and showing higher accuracy and robustness.

Contribution

The proposed LTC method provides a segmentation-free, more accurate approach for measuring cardiac strain from echocardiograms, validated on synthetic and clinical data.

Findings

Achieved higher correlation with ground truth (R^2=0.91 for strain) compared to conventional methods.

Unaffected by contrast-to-noise ratio, maintaining accuracy across different image qualities.

Improved clinical discrimination of cardiac function with AUC=0.85, surpassing existing algorithms.

Abstract

We present a new method for measuring global longitudinal strain and global longitudinal strain rate from 2D echocardiograms using a logarithmic-transform correlation (LTC) method. In contrast to traditional echocardiography strain analysis, our approach does not require cardiac chamber segmentation and regularization. The algorithm was benchmarked against two conventional strain analysis methods using synthetic left ventricle ultrasound images. Measurement error was assessed as a function of contrast-to-noise ratio (CNR) using mean absolute error and root-mean-square error and showed better agreement to the ground truth for strain($R^2$ = 0.91) and strain rate($R^2$ = 0.85) as compared to conventional algorithms (strain($R^2$ = 0.7), strain rate($R^2$ = 0.7)). Also, our method was unaffected by CNR. A 200% increase in strain measurement accuracy was observed compared to the conventional algorithms. Subsequently, we tested the method using a 54-subject clinical cohort (20 subjects diseased with cardiomyopathy, 34 healthy controls). Our method distinguished between normal and abnormal left ventricular function with an AUC = 0.85, a 10% improvement over the conventional GLS algorithms.