Inter-domain Alignment for Predicting High-Resolution Brain Networks Using Teacher-Student Learning

Accurate and automated super-resolution image synthesis is highly desired since it has the great potential to circumvent the need for acquiring high-cost medical scans and a time-consuming preprocessing pipeline of neuroimaging data. However, existing deep learning frameworks are solely designed to predict high-resolution (HR) image from a low-resolution (LR) one, which limits their generalization ability to brain graphs (i.e., connectomes). A small body of works has focused on superresolving brain graphs where the goal is to predict a HR graph from a single LR graph. Although promising, existing works mainly focus on superresolving graphs belonging to the same domain (e.g., functional), overlooking the domain fracture existing between multimodal brain data distributions (e.g., morphological and structural). To this aim, we propose a novel inter-domain adaptation framework namely, Learn to SuperResolve Brain Graphs with Knowledge Distillation Network (L2S-KDnet), which adopts a teacher-student paradigm to superresolve brain graphs. Our teacher network is a graph encoder-decoder that firstly learns the LR brain graph embeddings, and secondly learns how to align the resulting latent representations to the HR ground truth data distribution using an adversarial regularization. Ultimately, it decodes the HR graphs from the aligned embeddings. Next, our student network learns the knowledge of the aligned brain graphs as well as the topological structure of the predicted HR graphs transferred from the teacher. We further leverage the decoder of the teacher to optimize the student network. In such a way, we are not only bringing the learned embeddings from both networks closer to each other but also their predicted HR graphs. L2S-KDnet presents the first TS architecture tailored for brain graph super-resolution synthesis that is based on inter-domain alignment. Our experimental results demonstrate substantial performance gains over benchmark methods. Our code is available at https://github.com/basiralab/L2S-KDnet. ? corresponding author: irekik@itu.edu.tr, http://basira-lab.com. † : co-first authors. ar X iv :2 11 0. 03 45 2v 1 [ ee ss .I V ] 6 O ct 2 02 1

[1]  Martin G. Everett,et al.  A Graph-theoretic perspective on centrality , 2006, Soc. Networks.

[2]  Bo Zhou,et al.  Limited Angle Tomography Reconstruction: Synthetic Reconstruction via Unsupervised Sinogram Adaptation , 2019, IPMI.

[3]  Shuai Wang,et al.  Synthesized 7T MRI from 3T MRI via deep learning in spatial and wavelet domains , 2020, Medical Image Anal..

[4]  Islem Rekik,et al.  7 Years of Developing Seed Techniques for Alzheimer's Disease Diagnosis Using Brain Image and Connectivity Data Largely Bypassed Prediction for Prognosis , 2019, PRIME@MICCAI.

[5]  O. Sporns,et al.  Network neuroscience , 2017, Nature Neuroscience.

[6]  Islem Rekik,et al.  Graph Neural Networks in Network Neuroscience , 2021, ArXiv.

[7]  Amos Storkey,et al.  Advances in Neural Information Processing Systems 20 , 2007 .

[8]  Wen Li,et al.  Automated parcellation of the brain surface generated from magnetic resonance images , 2013, Front. Neuroinform..

[9]  Islem Rekik,et al.  Predicting High-Resolution Brain Networks Using Hierarchically Embedded and Aligned Multi-resolution Neighborhoods , 2019, PRIME@MICCAI.

[10]  Yan Shen,et al.  Brain Tumor Segmentation on MRI with Missing Modalities , 2019, IPMI.

[11]  Geoffrey E. Hinton,et al.  Distilling the Knowledge in a Neural Network , 2015, ArXiv.

[12]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[13]  Debiao Li,et al.  Efficient and Accurate MRI Super-Resolution using a Generative Adversarial Network and 3D Multi-Level Densely Connected Network , 2018, MICCAI.

[14]  Islem Rekik,et al.  Brain graph super-resolution for boosting neurological disorder diagnosis using unsupervised multi-topology connectional brain template learning , 2020, Medical Image Anal..

[15]  Georgios Tzimiropoulos,et al.  Knowledge distillation via softmax regression representation learning , 2021, ICLR.

[16]  Rajeev Motwani,et al.  The PageRank Citation Ranking : Bringing Order to the Web , 1999, WWW 1999.

[17]  Phillip Bonacich,et al.  Some unique properties of eigenvector centrality , 2007, Soc. Networks.

[18]  Verónica Vilaplana,et al.  Brain MRI super-resolution using 3D generative adversarial networks , 2018, ArXiv.

[19]  Islem Rekik,et al.  Brain Graph Super-Resolution Using Adversarial Graph Neural Network with Application to Functional Brain Connectivity , 2021, Medical Image Anal..

[20]  Max Welling,et al.  Semi-Supervised Classification with Graph Convolutional Networks , 2016, ICLR.

[21]  Graph Domain Adaptation with Localized Graph Signal Representations , 2019, ArXiv.

[22]  Islem Rekik,et al.  GSR-Net: Graph Super-Resolution Network for Predicting High-Resolution from Low-Resolution Functional Brain Connectomes , 2020, MLMI@MICCAI.

[23]  Pingkun Yan,et al.  Machine Learning in Medical Imaging: 10th International Workshop, MLMI 2019, Held in Conjunction with MICCAI 2019, Shenzhen, China, October 13, 2019, Proceedings , 2019, MLMI@MICCAI.

[24]  Elif Vural,et al.  Domain Adaptation on Graphs by Learning Aligned Graph Bases , 2018, IEEE Transactions on Knowledge and Data Engineering.