Related Experiment Video
Updated: Aug 13, 2025
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
42.7K
Fully Automatic Left Ventricle Segmentation Using Bilateral Lightweight Deep Neural Network.
Muhammad Ali Shoaib1,2, Joon Huang Chuah1, Raza Ali1,2
1Department of Electrical Engineering, Faculty of Engineering, Universiti Malaya, Kuala Lumpur 50603, Malaysia.
Life (Basel, Switzerland)
|January 21, 2023
Summary
This study introduces a fast and accurate deep learning model for segmenting the left ventricle (LV) in echocardiography images. The lightweight model improves upon existing methods, offering precise LV segmentation with reduced processing time.
Area of Science:
- Medical Imaging
- Cardiovascular Imaging
- Artificial Intelligence in Medicine
Background:
- Accurate left ventricle (LV) segmentation is crucial for cardiac quantitative analysis.
- Current semi-automatic methods are operator-dependent and subjective.
- Automatic LV segmentation from echocardiography is challenging due to image quality and boundary definition.
Purpose of the Study:
- To develop a single-stage, lightweight deep learning model for precise and rapid automatic segmentation of the LV in 2D echocardiography.
- To enhance segmentation accuracy and speed compared to existing state-of-the-art models.
Main Methods:
- A novel lightweight segmentation model utilizing a backbone network for feature extraction.
- Parallel spatial and channel feature units for feature enhancement.
- An integrated unit for merging refined features and performing LV segmentation.
- Comparison with DeepLab, FCN, and Mask RCNN models.
Main Results:
- The proposed model achieved a Dice similarity index of 0.9446, Intersection over Union of 0.8445, and accuracy of 0.9742.
- Demonstrated significantly reduced training and segmentation times compared to established models.
- Outperformed DeepLab, FCN, and Mask RCNN in quantitative segmentation performance and speed.
Conclusions:
- The developed lightweight model offers superior accuracy and speed for automatic LV segmentation in echocardiography.
- This approach addresses limitations of current methods, providing a more efficient and reliable tool for cardiac analysis.
- The model's performance indicates its potential for clinical application in quantitative cardiac assessments.
