Related Experiment Video
Updated: May 3, 2026
06:34
Evaluation of Left Ventricular Structure and Function using 3D Echocardiography
Published on: October 28, 2020
4.1K
Segmentation of 4D echocardiography using stochastic online dictionary learning.
Xiaojie Huang1, Donald P Dione2, Ben A Lin2
1Department of Electrical Engineering, Yale University, New Haven, CT, USA. xiaojie.huang@yale.edu
Medical Image Computing and Computer-Assisted Intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention
|February 8, 2014
Summary
This study introduces a novel stochastic online dictionary learning method for segmenting left ventricular borders in 4D echocardiography. The approach enhances accuracy, robustness, and speed compared to prior techniques.
Area of Science:
- Medical Imaging
- Biomedical Engineering
- Machine Learning
Background:
- Dictionary learning effectively utilizes spatiotemporal coherence for echocardiographic segmentation.
- Previous methods faced limitations in computational efficiency and adaptability.
Purpose of the Study:
- To present a stochastic online dictionary learning approach for segmenting left ventricular borders from 4D echocardiography.
- To improve upon existing methods by reducing memory and computational costs.
Main Methods:
- Employs stochastic approximations, processing data in mini-batches for efficiency.
- Optimizes dictionaries and weights by aggregating past information and adapting to dynamic data.
- Controls the update rate of past information based on appearance scale for balanced learning.
Main Results:
- Demonstrates higher accuracy and robustness in segmenting left ventricular borders.
- Achieves faster processing speeds compared to classical batch algorithms.
- Validated on 26 4D echocardiographic images.
Conclusions:
- The proposed stochastic online dictionary learning method offers a superior alternative for 4D echocardiographic segmentation.
- This technique enhances efficiency and adaptability in analyzing cardiac imaging data.
- The findings suggest significant advancements in automated cardiac segmentation accuracy and speed.
