Congratulations to MICCAI Society Young Scientist Award Recipients

Congratulations to Xiaohuan Cao (UNC IDEA Lab), winner of the MICCAI Society Young Scientist Award for her paper entitled “Learning-based Multimodal Image Registration for Prostate Cancer Radiation Therapy” [Xiaohuan Cao, Yaozong Gao, Jianhua Yang, Guorong Wu, Dinggang Shen].

Xiaohuan MICCAIA bi-directional image synthesis based non-rigid multimodal registration method is proposed to register pelvic MRI and CT for facilitating prostate cancer radiation therapy. Two main contributions in this paper: 1) To bridge the appearance gap between two modalities, the structured random forest with auto-context modal is used to perform bi-directional image synthesis: not only synthesizing CT from MRI, but also synthesizing MRI from CT. 2) An iterative dual-core deformation fusion framework is further proposed to drive the deformation pathway between MRI and CT by fully utilizing the complementary information from both modalities. The accurate experimental results show the high potential of the proposed method to transfer to routine radiation therapy.

Also, congratulations to Mingxia Liu (UNC IDEA Lab) for making the MICCAI Society Young Scientist Awards Runners up for her paper “Diagnosis of Alzheimer’s Disease Using View-Aligned Hypergraph Learning with Incomplete Multi-Modality Data” [Mingxia Liu, Jun Zhang, Pew-Thian Yap, Dinggang Shen].

Mingxia MICCAIExisting multi-view learning methods for diagnosis of Alzheimer’s disease (AD) usually ignore the underlying coherence among views, which may lead to suboptimal learning performance. In this paper, a view-aligned hypergraph learning method is proposed to explicitly model the coherence among different views. Specifically, the original data is first divided into several views based on possible combinations of modalities, followed by a sparse representation based hypergraph construction process in each view. A view-aligned hypergraph classification model is then proposed, by using a view-aligned regularizer to model the view coherence. Finally, the class probability scores are further assembled via a multi-view label fusion method to make a final classification decision. This method has been on the baseline ADNI-1 database with 807 subjects and three modalities (i.e., MRI, PET, and CSF). The proposed method achieves at least a 4.6% improvement in classification accuracy compared with state-of-the-art methods for AD diagnosis.

Additional information can be found at