Related Experiment Video
Updated: Nov 9, 2025
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
43.1K
LBTS-Net: A fast and accurate CNN model for brain tumour segmentation
Mohammed A M Abdullah1, Sinan Alkassar1, Bilal Jebur1
1Computer and Information Engineering Department Ninevah University Mosul Iraq.
Healthcare Technology Letters
|April 14, 2021
Summary
This study introduces a lightweight brain tumour segmentation network (LBTS-Net) for fast and accurate segmentation. Integrated transfer learning enhances robustness, achieving high accuracy on the BRATS2015 database.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neuroscience
Background:
- Accurate brain tumour segmentation is challenging due to complex tumour structures and irregular shapes.
- Existing methods may lack efficiency or robustness in segmenting brain tumours.
Purpose of the Study:
- To propose a lightweight brain tumour segmentation network (LBTS-Net) for efficient and accurate segmentation.
- To integrate transfer learning for enhanced robustness in brain tumour segmentation.
Main Methods:
- Developed LBTS-Net, a lightweight convolutional neural network based on VGG architecture.
- Reduced convolution filters and employed depth-wise convolution to optimize VGG-16 and VGG-19 networks.
- Utilized transfer learning and modified classification layers for fine-tuning.
Main Results:
- Achieved a global accuracy of 98.11% and a Dice score of 91% on the BRATS2015 database.
- Demonstrated significantly improved computational efficiency with approximately half the parameters of standard VGG networks.
- Confirmed robustness and state-of-the-art performance compared to existing methods.
Conclusions:
- LBTS-Net offers a computationally efficient and accurate solution for brain tumour segmentation.
- The integration of transfer learning contributes to a robust segmentation model.
- This work presents a novel, lightweight, and tailored network for brain tumour segmentation.
