Related Experiment Video
Updated: Sep 10, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
Multi-view parallel convolutional network for organ segmentation in mediastinal region on CT images
Yining Xie1, Wei Zhou1, Jiayi Ma2
1College of Mechanical and Electrical Engineering, Northeast Forestry University, Harbin, 150040, China.
Summary
This study introduces MVPCNet, a novel deep learning model for segmenting mediastinal organs in lung CT scans. MVPCNet achieves high accuracy and efficiency, outperforming existing methods for medical image segmentation.
Area of Science:
- Medical Imaging
- Computer Vision
- Artificial Intelligence
Background:
- Mediastinal organ segmentation in lung CT is vital for precise localization.
- Existing methods struggle with complex organ topologies, morphological variations, and feature differentiation.
Purpose of the Study:
- To develop a novel deep learning network, MVPCNet, for accurate and efficient mediastinal organ segmentation.
- To address limitations in current medical image segmentation techniques.
Main Methods:
- Proposed a multi-view parallel convolutional network (MVPCNet) with an efficient U-shaped encoder-decoder framework.
- Incorporated multi-view parallel convolution modules (MVPM) and dual-path backbone structures (DPBS) for feature extraction.
- Utilized efficient dual-channel bottleneck structures (EDC-BS) and region fusion small-kernel deformable attention (RF-SKDA).
Main Results:
- MVPCNet achieved an average Dice Coefficient of 90.59% and mIoU of 82.80% on a mediastinal organ dataset.
- The model demonstrated superior performance compared to advanced medical segmentation algorithms and lightweight semantic segmentation models.
- Achieved a compact parameter size of only 8.21 MB.
Conclusions:
- MVPCNet offers a highly accurate and computationally efficient solution for mediastinal organ segmentation in lung CT images.
- The proposed network effectively models complex topological structures and morphological variations.
- MVPCNet represents a significant advancement in medical image segmentation technology.
