Related Experiment Video
Updated: Aug 1, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.8K
An early detection and segmentation of Brain Tumor using Deep Neural Network
Mukul Aggarwal1, Amod Kumar Tiwari2, M Partha Sarathi3
1Dr. A.P.J. Abdul Kalam Technical University, Lucknow, Uttar Pradesh, India.
BMC Medical Informatics and Decision Making
|April 26, 2023
Summary
This study introduces an improved Residual Network (ResNet) for brain tumor segmentation in MRI scans. The enhanced ResNet accelerates learning and improves accuracy for better tumor diagnosis and treatment planning.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neuroscience
Background:
- Brain tumor segmentation using Magnetic Resonance Imaging (MRI) is vital for diagnosis, prognosis, and treatment planning.
- Tumor variability in shape, size, and appearance poses significant segmentation challenges.
- Deep Neural Networks (DNNs) show promise but face training difficulties like gradient issues.
Purpose of the Study:
- To develop an efficient brain tumor segmentation method addressing DNN gradient issues.
- To enhance the ResNet architecture for improved precision and faster learning.
Main Methods:
- An improved Residual Network (ResNet) was developed, focusing on information flow, residual blocks, and projection shortcuts.
- The methodology optimizes computational costs and accelerates the segmentation process.
- The improved ResNet maintains detailed information across network layers.
Main Results:
- The proposed improved ResNet demonstrated enhanced precision and faster learning.
- Computational costs were minimized, and the learning process was accelerated.
- The network effectively addressed information flow, residual blocks, and projection shortcuts.
Conclusions:
- The improved ResNet offers an efficient solution for brain tumor segmentation.
- This method overcomes limitations of traditional DNNs and improves segmentation accuracy.
- Experimental results show over 10% improvement in accuracy, recall, and f-measure compared to CNN and FCN.
