Computer Vision News - November 2020

2 Young Scientist Winner 2 Best of MICCAI 2020 Lu Zhang is a third-year PhD student at the University of Texas at Arlington in the Computer Science and Engineering department, under the supervision of Dr Dajiang Zhu. Fresh from winning a prestigious Young Scientist award at MICCAI 2020, she speaks to us about her work on the brain. Lu Zhang is a third-year PhD student at the University of Texas at Arlington in the Computer Science and Engineering department, under the supervision of Dr Dajiang Zhu. Fresh from winning a prestigious Young Scientist award at MICCAI 2020, she speaks to us about her work on the brain. This work proposes an efficient method to model the complex relationship between brain structure and function , both of which are critical for understanding the organizational principles of the human brain and gaining knowledge for treating many different brain diseases. This study can be challenging for two reasons. The many-to-one function- structure mode of the brain means that although brain structure is stable in the short term, its function signal can be diverse. Also, there are many direct or indirect relationships in both structural and functional connectivity . “We proposed a generative adversarial network (GAN) to solve the brain’s many-to-one function-structure mode because it learns distribution to another distribution,” Lu explains. “However, the GAN is unstable when training, so we designed a novel structure-preserving loss function composed of three parts: basic GAN loss , MSE loss , and PCC loss . The model creates a predicted structure and by comparing the predicted one to the real one, it can learn the relationship.” The GAN loss is used to learn from the data because the exact relationship Recovering Brain Structural Connectivity fromFunctional ConnectivityviaMulti-GCN Based Generative Adversarial Network