AUTHOR=Seong Si-Baek , Pae Chongwon , Park Hae-Jeong TITLE=Geometric Convolutional Neural Network for Analyzing Surface-Based Neuroimaging Data JOURNAL=Frontiers in Neuroinformatics VOLUME=Volume 12 - 2018 YEAR=2018 URL=https://www.frontiersin.org/journals/neuroinformatics/articles/10.3389/fninf.2018.00042 DOI=10.3389/fninf.2018.00042 ISSN=1662-5196 ABSTRACT=In the machine learning, one of the most popular deep learning methods is convolutional neural network (CNN), which utilizes shared local filters and hierarchical information processing analogous to the brain’s visual system. Despite its popularity in recognizing two-dimensional images, the conventional CNN is not directly applicable to semi-regular geometric mesh surfaces, over which the cerebral cortex is often represented. In order to apply CNN to surface-based brain research, we propose a geometric CNN (gCNN) that deals with data representation over a mesh surface and renders pattern recognition in a multi-shell mesh structure. To make it compatible with the conventional CNN toolbox, gCNN includes data sampling over the surface and a data reshaping method for the convolution and pooling layers. We evaluated the performance of gCNN in sex classification using the cortical thickness maps of both hemispheres in the Human Connectome Project. The classification accuracy of gCNN was significantly higher than those of a support vector machine and a two-dimensional CNN for thickness maps generated by a map projection. gCNN also demonstrated position invariance of local features that renders reuse of its pre-trained model for applications other than the purpose for which the model was trained, without significant distortions in the final outcome. The superior performance of gCNN is attributable to the CNN properties stemming from its brain-like architecture and surface-based representation of the cortical information. gCNN provides much-needed access to surface-based machine learning that can be used in both scientific investigations and clinical applications.