RODEO: Robust DE-aliasing autoencOder for real-time medical image reconstruction

详细信息    查看全文

• 作者：Janki MehtaAuthor Vitae ; Angshul MajumdarAuthor Vitae
• 关键词：Autoencoder ; MRI ; Compressed sensing ; CT reconstruction
• 刊名：Pattern Recognition
• 出版年：2017
• 出版时间：March 2017
• 年：2017
• 卷：63
• 期：Complete
• 页码：499-510
• 全文大小：1620 K
• 卷排序：63

文摘

We address the problem of real-time reconstruction of MRI and CT images from sub-sampled data. Compressed Sensing is the de facto standard for solving such problems. This work proposes an alternate approach; instead of designing/formulating the inversion, we ‘learn’ it. Robust autoencoder has been proposed in this work, to learn the inversion. Results show, our method is only slightly worse than CS methods but is capable of achieving real-time speed.