RDCU-Net: A Multi-Scale Residual Dilated Convolution U-Net with Spatial Pyramid Pooling for Brain Tumor Segmentation

Tumors refer to abnormal growth of cells in the body. Early diagnosis of tumors plays a crucial role in improving treatment conditions , quality of life and patient survival. Deep learning methods are effective for medical image segmentation, but they struggle with tumors in magnetic resonance image...

Full description

Saved in:
Bibliographic Details
Main Authors: Mohammad Soltani-Gol, Akbar Asgharzadeh-Bonab, Hamid Soltanian-Zadeh, Jalil Mazlum
Format: Article
Language:English
Published: Amirkabir University of Technology 2024-03-01
Series:AUT Journal of Electrical Engineering
Subjects:
Online Access:https://eej.aut.ac.ir/article_5325_d173565445d5c1945e4b83ac88d62f56.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849434195698384896
author Mohammad Soltani-Gol
Akbar Asgharzadeh-Bonab
Hamid Soltanian-Zadeh
Jalil Mazlum
author_facet Mohammad Soltani-Gol
Akbar Asgharzadeh-Bonab
Hamid Soltanian-Zadeh
Jalil Mazlum
author_sort Mohammad Soltani-Gol
collection DOAJ
description Tumors refer to abnormal growth of cells in the body. Early diagnosis of tumors plays a crucial role in improving treatment conditions , quality of life and patient survival. Deep learning methods are effective for medical image segmentation, but they struggle with tumors in magnetic resonance images (MRI) due to variations in intensity and appearance. Existing models like U-Net face challenges due to the integration of high-level and low-level features, leading to confusion. Our proposed model addresses the above issues by utilizing two techniques and fewer parameters compared to the existing methods, achieving higher accuracy. In the first technique, dilated convolution (DC) blocks with proportional rates are used to integrate high-level and low-level features. The second technique involves selecting dilated spatial pyramid (DSP) blocks, which increase the receptive field of features while maintaining their resolution, contributing to the network's generalization. The proposed model improves training, network depth, and feature extraction by incorporating a residual block. It outperforms the traditional U-Net model in terms of segmentation accuracy and network stability. We evaluated the model using the BraTS 2018 dataset, obtaining Dice coefficients of 0.906, 0.817, and 0.839 for the whole tumor (WT), the enhancing tumor (ET), and the tumor core (TC), respectively.
format Article
id doaj-art-6fc5df34706e49d19c134018c2db416d
institution Kabale University
issn 2588-2910
2588-2929
language English
publishDate 2024-03-01
publisher Amirkabir University of Technology
record_format Article
series AUT Journal of Electrical Engineering
spelling doaj-art-6fc5df34706e49d19c134018c2db416d2025-08-20T03:26:44ZengAmirkabir University of TechnologyAUT Journal of Electrical Engineering2588-29102588-29292024-03-0156220321210.22060/eej.2023.22395.55385325RDCU-Net: A Multi-Scale Residual Dilated Convolution U-Net with Spatial Pyramid Pooling for Brain Tumor SegmentationMohammad Soltani-Gol0Akbar Asgharzadeh-Bonab1Hamid Soltanian-Zadeh2Jalil Mazlum3Department of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, IranDepartment of Electrical and Computer Engineering, College of Engineering, Urmia University, Urmia, IranDepartment of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, IranDepartment of Electrical Engineering, Shahid Beheshti Aeronautical University of Science and Technology, Tehran, IranTumors refer to abnormal growth of cells in the body. Early diagnosis of tumors plays a crucial role in improving treatment conditions , quality of life and patient survival. Deep learning methods are effective for medical image segmentation, but they struggle with tumors in magnetic resonance images (MRI) due to variations in intensity and appearance. Existing models like U-Net face challenges due to the integration of high-level and low-level features, leading to confusion. Our proposed model addresses the above issues by utilizing two techniques and fewer parameters compared to the existing methods, achieving higher accuracy. In the first technique, dilated convolution (DC) blocks with proportional rates are used to integrate high-level and low-level features. The second technique involves selecting dilated spatial pyramid (DSP) blocks, which increase the receptive field of features while maintaining their resolution, contributing to the network's generalization. The proposed model improves training, network depth, and feature extraction by incorporating a residual block. It outperforms the traditional U-Net model in terms of segmentation accuracy and network stability. We evaluated the model using the BraTS 2018 dataset, obtaining Dice coefficients of 0.906, 0.817, and 0.839 for the whole tumor (WT), the enhancing tumor (ET), and the tumor core (TC), respectively.https://eej.aut.ac.ir/article_5325_d173565445d5c1945e4b83ac88d62f56.pdfimage segmentationdeep convolutional neural networkmagnetic resonance imaging (mri)brain tumor
spellingShingle Mohammad Soltani-Gol
Akbar Asgharzadeh-Bonab
Hamid Soltanian-Zadeh
Jalil Mazlum
RDCU-Net: A Multi-Scale Residual Dilated Convolution U-Net with Spatial Pyramid Pooling for Brain Tumor Segmentation
AUT Journal of Electrical Engineering
image segmentation
deep convolutional neural network
magnetic resonance imaging (mri)
brain tumor
title RDCU-Net: A Multi-Scale Residual Dilated Convolution U-Net with Spatial Pyramid Pooling for Brain Tumor Segmentation
title_full RDCU-Net: A Multi-Scale Residual Dilated Convolution U-Net with Spatial Pyramid Pooling for Brain Tumor Segmentation
title_fullStr RDCU-Net: A Multi-Scale Residual Dilated Convolution U-Net with Spatial Pyramid Pooling for Brain Tumor Segmentation
title_full_unstemmed RDCU-Net: A Multi-Scale Residual Dilated Convolution U-Net with Spatial Pyramid Pooling for Brain Tumor Segmentation
title_short RDCU-Net: A Multi-Scale Residual Dilated Convolution U-Net with Spatial Pyramid Pooling for Brain Tumor Segmentation
title_sort rdcu net a multi scale residual dilated convolution u net with spatial pyramid pooling for brain tumor segmentation
topic image segmentation
deep convolutional neural network
magnetic resonance imaging (mri)
brain tumor
url https://eej.aut.ac.ir/article_5325_d173565445d5c1945e4b83ac88d62f56.pdf
work_keys_str_mv AT mohammadsoltanigol rdcunetamultiscaleresidualdilatedconvolutionunetwithspatialpyramidpoolingforbraintumorsegmentation
AT akbarasgharzadehbonab rdcunetamultiscaleresidualdilatedconvolutionunetwithspatialpyramidpoolingforbraintumorsegmentation
AT hamidsoltanianzadeh rdcunetamultiscaleresidualdilatedconvolutionunetwithspatialpyramidpoolingforbraintumorsegmentation
AT jalilmazlum rdcunetamultiscaleresidualdilatedconvolutionunetwithspatialpyramidpoolingforbraintumorsegmentation