Related Experiment Video
Updated: Jun 14, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.7K
Brain tumor image segmentation method using hybrid attention module and improved mask RCNN
1School of Applied Science, Macao Polytechnic University, Macau, 999078, China. p2316169@mpu.edu.mo.
Scientific Reports
|September 4, 2024
Summary
This study enhances brain tumor segmentation in MRI using an improved mask region-based convolutional neural network with attention mechanisms. The new model offers more precise tumor localization for medical analysis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neuro-oncology
Background:
- Automated analysis of brain tumor magnetic resonance imaging (MRI) is crucial for efficient medical diagnosis.
- Existing methods may lack the precision required for detailed tumor characterization.
Purpose of the Study:
- To develop an enhanced instance segmentation method for precise brain tumor segmentation in MRI.
- To improve the accuracy and efficiency of automated brain tumor analysis.
Main Methods:
- An enhanced mask region-based convolutional neural network (R-CNN) was developed.
- Incorporated squeeze-and-excitation networks (channel attention) and concatenated attention neural network (spatial attention).
- Utilized Residual Network-50 with an attention module and feature pyramid network as the backbone, alongside region proposal network and region of interest alignment.
Main Results:
- The enhanced model achieved a precision of 90.72% (0.76% increase).
- Recall was 91.68% (0.95% increase), and Mean Intersection over Union (IoU) was 94.56% (1.39% increase).
- Demonstrated improved feature extraction efficiency for brain tumors.
Conclusions:
- The proposed method achieves precise segmentation of brain tumors in MRI.
- Facilitates accurate measurement of tumor dimensions, aiding in comprehensive diagnostic information for clinicians.
