Related Experiment Video
Updated: Dec 8, 2025
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
875
A novel solution of using deep learning for left ventricle detection: Enhanced feature extraction
Kiran Sharma1, Abeer Alsadoon1, P W C Prasad1
1School of Computing and Mathematics, Charles Sturt University, Sydney Campus, Australia.
Computer Methods and Programs in Biomedicine
|September 21, 2020
Summary
This study introduces an enhanced deep learning model for accurate left ventricle (LV) detection in echocardiograms, significantly reducing processing time and overcoming common deep learning challenges.
Area of Science:
- Cardiovascular Imaging
- Artificial Intelligence in Medicine
- Medical Image Analysis
Background:
- Deep learning models face challenges like overfitting and vanishing gradients in echocardiographic left ventricle (LV) detection.
- Existing methods struggle with accuracy and processing speed for LV analysis.
Purpose of the Study:
- To enhance the accuracy and reduce processing time for left ventricle (LV) detection in echocardiographic images.
- To mitigate overfitting and vanishing gradient issues in deep learning models for cardiac imaging.
Main Methods:
- An enhanced deep convolutional neural network (CNN) incorporating an additional convolutional layer and dropout layer was developed.
- Data augmentation techniques were employed to improve feature extraction accuracy for LV detection.
Main Results:
- The model achieved 94% accuracy in left ventricle (LV) detection across four diverse pathological groups (heart failure with/without infarction, hypertrophy, healthy).
- Average processing time was reduced from 0.45s to 0.34s, demonstrating improved efficiency.
Conclusions:
- The proposed system effectively enhances accuracy and decreases processing time for left ventricle (LV) detection.
- The research successfully addresses and resolves overfitting issues in deep learning models applied to cardiac image analysis.
