Related Experiment Video
Updated: Jun 23, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.7K
An Efficient Muscle Segmentation Method via Bayesian Fusion of Probabilistic Shape Modeling and Deep Edge Detection
IEEE Transactions on Bio-Medical Engineering
|June 18, 2024
Summary
This study introduces a new method for segmenting paraspinal muscles from MRI scans, improving accuracy even with limited data. This aids in assessing low back pain more effectively.
Area of Science:
- Medical Imaging
- Biomedical Engineering
- Computational Anatomy
Background:
- Accurate paraspinal muscle segmentation from MRI is vital for quantitative assessment of low back pain.
- Current methods struggle with unclear boundaries and shape variations, especially with limited training data.
Purpose of the Study:
- To develop a novel approach for paraspinal muscle shape modeling and inference.
- To enhance segmentation accuracy and efficiency using minimal training data.
Main Methods:
- A probabilistic shape model (PSM) using Fourier basis functions and Gaussian processes (GPs) to encode 3D muscle shapes.
- A Bayesian framework integrating the PSM prior with deep-learning-based edge detections and sparse manual annotations.
Main Results:
- Achieved Dice similarity coefficient exceeding 90% with only three annotated slices.
- Demonstrated superior performance compared to existing methods on public and clinical datasets.
- Requires small training datasets and offers rapid inference speeds.
Conclusions:
- The proposed method enables precise 2D and 3D paraspinal muscle assessment.
- Facilitates better understanding of muscle changes in various conditions.
- Potential to enhance treatment outcomes for low back pain.
