Mechanical Performance Separation of Cardiac by Nonlinear Processing of Ultrasound B-Mode Images

Hamidreza Fazilatnezhad; Parisa  Rangraz; Fereidon  Noshirvan Rahatabad

doi:10.18502/fbt.v10i1.11514

Hamidreza Fazilatnezhad Department of Biomedical Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran
Parisa Rangraz Department of Biomedical Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran
Fereidon Noshirvan Rahatabad Department of Biomedical Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran

DOI: https://doi.org/10.18502/fbt.v10i1.11514

Keywords: Ultrasound; Left Ventricular Ejection Fraction; Nonlinear Analysis; Statistical Analysis; Image Segmentation; Feature Extraction

Abstract

Purpose: Accurate measurement of Left Ventricular Ejection Fraction (LVEF) is critical for diagnosis of and predicting Left Ventricular (LV) arrhythmias. This study aims to estimate LVEF using nonlinear and statistical analysis in echocardiography images.

Materials and Methods: The Cardiac Acquisition for Multi-Structure Ultrasound Segmentation (CAMUS) dataset is used to estimate LVEF. This dataset includes ultrasound images of 60 patients in two different groups (LVEF > 55%, LVEF < 45%). Region growing technique and Anatomical markers were used for segmentation of LV in images to measure region changes. LV region changes were investigated using nonlinear and statistical analysis. To facilitate estimating LVEF, feature extraction and Artificial Neural Networks (ANN) have been used.

Results: The results show that the changes in the LV region in LVEF < 45% have a mean value of 3.254 while LVEF > 55% has a lower mean value of 3.071, but the mean of variance is 3.818 while for LVEF < 45% is 3.471 which can be concluded that the data scatter in LVEF > 55% was higher than the mean and indicates more significant changes in the LV region.

Conclusion: LVEF estimated using nonlinear and statistical analysis shows a Mean Square Error (MSE) of 5.15.