An innovative image segmentation approach for brain tumor based on 3D-Pix2Pix

The brain tumor image segmentation method based on deep learning is an important technical means to improve the performance of brain tumor diagnosis. GAN can solve the problem of manual labeling difficulties and strong subjectivity in image processing. Therefore, a semi-supervised image segmentation...

Full description

Saved in:
Bibliographic Details
Published in2021 6th International Symposium on Computer and Information Processing Technology (ISCIPT) pp. 542 - 545
Main Authors Li, Mengxin, Zhang, Tianhui, Li, Songang
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.06.2021
Subjects
Brain modeling
brain tumor segmentation
CGAN
Convolution
Generative adversarial networks
Generators
Image segmentation
Manuals
pix2pix
Training
u-net
Online AccessGet full text
DOI10.1109/ISCIPT53667.2021.00115

Cover

More Information
Summary:The brain tumor image segmentation method based on deep learning is an important technical means to improve the performance of brain tumor diagnosis. GAN can solve the problem of manual labeling difficulties and strong subjectivity in image processing. Therefore, a semi-supervised image segmentation method for brain tumors based on 3D-Pix2Pix is proposed on the basis of CGAN. The generator adopts the U-Net network combined with the residual attention module to avoid gradient disappearance and over-fitting problems in training, and increase the dropout regularization to improve the generalization ability of the model. Experimental results show that this method basically achieves accurate segmentation of brain tumors.
DOI:10.1109/ISCIPT53667.2021.00115