Magnetic Resonance Imaging Brain Tumor Image Segmentation Based on Improved U-Net

Haiwei Mu; Ying Guo; Xinghui Quan; Zhimin Cao; Jian Han

doi:10.3788/LOP202158.0410022

Journals >Laser & Optoelectronics Progress >Volume 58 >Issue 4 >Page 0410022 > Article

Laser & Optoelectronics Progress
Vol. 58, Issue 4, 0410022 (2021)

Magnetic Resonance Imaging Brain Tumor Image Segmentation Based on Improved U-Net

Haiwei Mu^1、2, Ying Guo^1、2, Xinghui Quan^1、2、*, Zhimin Cao^1、2, and Jian Han^1、2

Author Affiliations

¹School of Physics and Electrical Engineering, Northeast Petroleum University, Daqing, Heilongjiang 163318, China

²Research and Development Center for Testing and Measurement Technology and Instrumentation, Heilongjiang Province Universities, Northeast Petroleum University, Daqing, Heilongjiang 163318, China

show less

DOI: 10.3788/LOP202158.0410022 Cite this Article Set citation alerts

Haiwei Mu, Ying Guo, Xinghui Quan, Zhimin Cao, Jian Han. Magnetic Resonance Imaging Brain Tumor Image Segmentation Based on Improved U-Net[J]. Laser & Optoelectronics Progress, 2021, 58(4): 0410022 Copy Citation Text

show less

Fig. 1. Structure diagram of the dense jump connection

Download full size

Fig. 2. Structure diagram of the improved residual module

Download full size

Fig. 3. Structure of the FRN layer

Download full size

Fig. 4. Structure of the improved network

Download full size

Fig. 5. MRI image of brain tumor. (a) FLAIR; (b) T1; (c) T1ce; (d) T2; (e) GT

Download full size

Fig. 6. Segmentation results of MRI images by different algorithms. (a) Patient1--(g) patient7

Download full size

Algorithm	Dice			PPV			Sensitivity			Hausdorff
Algorithm	WT	TC	ET	WT	TC	ET	WT	TC	ET	WT	TC	ET
FCN8	0.81	0.80	0.70	0.79	0.82	0.69	0.85	0.88	0.75	2.98	1.85	3.20
U-Net	0.83	0.81	0.76	0.85	0.84	0.77	0.86	0.89	0.82	2.66	1.75	2.84
ResNet	0.84	0.82	0.77	0.86	0.85	0.79	0.86	0.90	0.82	2.63	1.72	2.81
Ours	0.88	0.84	0.80	0.90	0.86	0.82	0.89	0.91	0.84	2.52	1.51	2.66

Table 1. Tumor segmentation results of the 4 models

Algorithm	Dice			Hausdorff
Algorithm	WT	TC	ET	WT	TC	ET
Ref. [27]	0.7385	0.6931	0.5772	3.4852	2.2848	3.5821
Ref. [28]	0.8106	0.7823	0.7017	3.0646	2.0548	3.1425
Ref. [29]	0.8335	0.8051	0.7634	2.8368	1.8267	2.8459
Ours	0.8843	0.8402	0.8098	2.5252	1.5144	2.6616

Table 2. Tumor segmentation results of different algorithms

Haiwei Mu, Ying Guo, Xinghui Quan, Zhimin Cao, Jian Han. Magnetic Resonance Imaging Brain Tumor Image Segmentation Based on Improved U-Net[J]. Laser & Optoelectronics Progress, 2021, 58(4): 0410022

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information