Related Experiment Video
Updated: Nov 4, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.1K
CLCU-Net: Cross-level connected U-shaped network with selective feature aggregation attention module for brain tumor
1School of Control Science and Engineering, Shandong University, Jinan 250061, China.
Computer Methods and Programs in Biomedicine
|May 25, 2021
Summary
This study introduces a new deep learning network for precise brain tumor segmentation. The novel approach effectively utilizes multi-scale features, significantly improving diagnostic accuracy for clinicians.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Brain tumors represent a critical global health challenge.
- Deep convolutional neural networks have advanced brain tumor segmentation.
- Existing methods often underutilize multi-scale features, limiting accuracy.
Purpose of the Study:
- To develop an advanced deep learning network for enhanced brain tumor segmentation.
- To improve the extraction and utilization of multi-scale features in segmentation tasks.
- To provide clinicians with a more accurate tool for brain tumor diagnosis and surgical planning.
Main Methods:
- Proposed a novel cross-level connected U-shaped network (CLCU-Net).
- Introduced a Segmented Attention Module (SAM) for selective feature aggregation.
- Employed deep supervision and spatial pyramid pooling (SSP) to optimize performance.
Main Results:
- Achieved a Dice Score of 88.5% on the BRATS 2018 dataset for whole tumor segmentation.
- Outperformed six state-of-the-art methods in segmentation accuracy.
- Demonstrated superior suitability of the proposed attention module for segmentation tasks.
Conclusions:
- The CLCU-Net with SAM effectively segments brain tumors.
- The method shows significant potential for clinical practice implementation.
- Accurate segmentation aids in improved patient diagnosis and treatment planning.
