Related Experiment Video
Updated: Aug 26, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
Convolutional bi-directional learning and spatial enhanced attentions for lung tumor segmentation
Ping Xuan1, Bin Jiang2, Hui Cui3
1School of Computer Science and Technology, Heilongjiang University, Harbin, China; Department of Computer Science, School of Engineering, Shantou University, Shantou, China.
Computer Methods and Programs in Biomedicine
|October 7, 2022
Summary
This study introduces PRCS, a novel method for accurate lung tumor segmentation in CT scans. The approach enhances learning of contextual and spatial information, improving segmentation for challenging tumors.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Accurate lung tumor segmentation in CT scans is challenging due to tumor variability.
- Learning semantic and spatial relationships is critical for precise segmentation.
Purpose of the Study:
- To develop an advanced segmentation method for lung tumors.
- To improve the learning of multi-channel contextual and spatial dependencies.
Main Methods:
- Proposed PRCS method integrating multi-channel contextual relations and spatial dependencies.
- Utilized convolutional bi-directional gated recurrent units for feature channel context.
- Implemented cross-channel region-level attention and position-enhanced self-attention mechanisms.
Main Results:
- PRCS outperformed seven state-of-the-art methods on lung tumor datasets.
- Demonstrated superior spatial overlapping and shape similarity in segmentation.
- Ablation studies confirmed the effectiveness of proposed innovations and generalization.
Conclusions:
- PRCS improves lung tumor segmentation by enhancing contextual and spatial relation learning.
- The method effectively handles variations and indistinct boundaries in tumors.
- PRCS offers an automated solution for lung cancer diagnosis and treatment planning.
