Back

Reproducibility of different methods of measuring left ventricular ejection fraction in echocardiography

Kawagishi, I.; Hashimoto, M.; Endo, Y.; Ono, S.; Takamuku, M.; Ohnishi, K.

2024-10-10 cardiovascular medicine
10.1101/2024.10.10.24315216 medRxiv
Show abstract

BackgroundLeft ventricular ejection fraction (LVEF) measurements by echocardiography are subject to problems such as low interobserver reproducibility, which depends on the skill of the observer, and variations in measurement technique. ObjectiveWe investigated the reproducibility of echocardiography by comparing the measurement results of a certified sonographer with those of control students. MethodsA total of 30 subjects, 22 adult students and 8 staff members of Shubun University, were included in the study. Measurements were performed according to the standard protocols of the Japanese Society of Echocardiography. ResultsLVEF values were significantly lower with the modified Simpson method than with the Teichholz method (M-mode and B-mode) for both the sonographer and students; there was no difference between the M-mode and B-mode approaches for the sonographer, but B-mode was significantly lower for students. Comparing the sonographer and students, there was no significant difference with the M-mode approach, but students had significantly lower values with the B-mode and modified Simpson methods. Although the modified Simpson method is currently recommended for LVEF measurement, the lower values obtained by students using this method are primarily attributed to their inexperience with the measurement technique. ConclusionsIn the future, the evaluation of cardiac function by echocardiography will require reproducibility that does not rely on the skill of the observer, and the adoption of echocardiography methods incorporating artificial intelligence is anticipated.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.