Generative Adversarial Network for Medical Images (MI-GAN)

被引:155
作者
Iqbal, Talha [1 ]
Ali, Hazrat [1 ]
机构
[1] COMSATS Univ Islamabad, Dept Elect Engn, Abbottabad Campus, Islamabad, Pakistan
关键词
GAN; Medical imaging; Style transfer; Deep learning; Retinal images; SELECTION;
D O I
10.1007/s10916-018-1072-9
中图分类号
R19 [保健组织与事业(卫生事业管理)];
学科分类号
摘要
Deep learning algorithms produces state-of-the-art results for different machine learning and computer vision tasks. To perform well on a given task, these algorithms require large dataset for training However, deep learning algorithms lack generalization and suffer from over-fitting whenever trained on small dataset, especially when one is dealing with medical images. For supervised image analysis in medical imaging, having image data along with their corresponding annotated ground-truths is costly as well as time consuming since annotations of the data is done by medical experts manually. In this paper, we propose a new Generative Adversarial Network for Medical Imaging (MI-GAN). The MI-GAN generates synthetic medical images and their segmented masks, which can then be used for the application of supervised analysis of medical images. Particularly, we present MI-GAN for synthesis of retinal images. The proposed method generates precise segmented images better than the existing techniques. The proposed model achieves a dice coefficient of 0.837 on STARE dataset and 0.832 on DRIVE dataset which is state-of-the-art performance on both the datasets.
引用
收藏
页数:11
相关论文
共 42 条
[1]  
[Anonymous], ARXIV170609318
[2]  
[Anonymous], 2016, P ICLR
[3]  
[Anonymous], ARXIV150806576
[4]  
[Anonymous], 2017, PROC CVPR IEEE, DOI DOI 10.1109/CVPR.2017.632
[5]  
[Anonymous], 2016, P 33 INT C INT C MAC
[6]  
[Anonymous], 2017, IEEE I CONF COMP VIS, DOI DOI 10.1109/ICCV.2017.244
[7]  
[Anonymous], 2015, CoRR
[8]  
[Anonymous], ARXIV170602185
[9]  
[Anonymous], 2016, MED IMAGING 2016 IMA
[10]  
[Anonymous], 2016, P ADV NEUR INF PROC