斑点图案
生物信息学
帧(网络)
跟踪(教育)
拉伤
帧速率
人工智能
数学
计算机科学
计算机视觉
生物
解剖
遗传学
心理学
电信
基因
教育学
作者
Assami Rösner,Daniel Barbosa,Erling Aarsæther,Didrik Kjønås,Henrik Schirmer,Jan D'hooge
出处
期刊:European Journal of Echocardiography
[Oxford University Press]
日期:2015-03-09
卷期号:16 (10): 1137-1147
被引量:81
摘要
Ultrasound-derived myocardial strain can render valuable diagnostic and prognostic information. However, acquisition settings can have an important impact on the measurements. Frame rate (i.e. temporal resolution) seems to be of particular importance. The aim of this study was to find the optimal range of frame rates needed for most accurate and reproducible 2D strain measurements using a 2D speckle-tracking software package. Synthetic two dimensional (2D) ultrasound grey-scale images of the left ventricle (LV) were generated in which the strain in longitudinal, circumferential, and radial direction were precisely known from the underlying kinematic LV model. Four different models were generated at frame rates between 20 and 110 Hz. The resulting images were repeatedly analysed. Results of the synthetic data were validated in 66 patients, where long- and short-axis recordings at different frame rates were analysed. In simulated data, accurate strain estimates could be achieved at >30 frames per cycle (FpC) for longitudinal and circumferential strains. Lower FpC underestimated strain systematically. Radial strain estimates were less accurate and less reproducible. Patient strain displayed the same plateaus as in the synthetic models. Higher noise and the presence of artefacts in patient data were followed by higher measurement variability. Standard machine settings with a FR of 50–60 Hz allow correct assessment of peak global longitudinal and circumferential strain. Correct definition of the region of interest within the myocardium as well as the reduction of noise and artefacts seem to be of highest importance for accurate 2D strain estimation.
科研通智能强力驱动
Strongly Powered by AbleSci AI