A Few-Shot Learning Graph Multi-trajectory Evolution Network for Forecasting Multimodal Baby Connectivity Development from a Baseline Timepoint

Charting the baby connectome evolution trajectory during the first year after birth plays a vital role in understanding dynamic connectivity development of baby brains. Such analysis requires acquisition of longitudinal connectomic datasets. However, both neonatal and postnatal scans are rarely acquired due to various difficulties. A small body of works has focused on predicting baby brain evolution trajectory from a neonatal brain connectome derived from a single modality. Although promising, large training datasets are essential to boost model learning and to generalize to a multi-trajectory prediction from different modalities (i.e., functional and morphological connectomes). Here, we unprecedentedly explore the question: “Can we design a few-shot learningbased framework for predicting brain graph trajectories across different modalities?” To this aim, we propose a Graph Multi-Trajectory Evolution Network (GmTE-Net), which adopts a teacher-student paradigm where the teacher network learns on pure neonatal brain graphs and the student network learns on simulated brain graphs given a set of different timepoints. To the best of our knowledge, this is the first teacherstudent architecture tailored for brain graph multi-trajectory growth prediction that is based on few-shot learning and generalized to graph neural networks (GNNs). To boost the performance of the student network, we introduce a local topology-aware distillation loss that forces ? corresponding author: irekik@itu.edu.tr, http://basira-lab.com. ar X iv :2 11 0. 03 53 5v 1 [ qbi o. N C ] 6 O ct 2 02 1 the predicted graph topology of the student network to be consistent with the teacher network. Experimental results demonstrate substantial performance gains over benchmark methods. Hence, our GmTENet can be leveraged to predict atypical brain connectivity trajectory evolution across various modalities. Our code is available at https: //github.com/basiralab/GmTE-Net.

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

[2]  Ievgen Redko,et al.  A survey on domain adaptation theory , 2020, ArXiv.

[3]  Islem Rekik,et al.  Learning-Guided Infinite Network Atlas Selection for Predicting Longitudinal Brain Network Evolution from a Single Observation , 2019, MICCAI.

[4]  Vishal M. Patel,et al.  Few Is Enough: Task-Augmented Active Meta-Learning for Brain Cell Classification , 2020, MICCAI.

[5]  M. Breakspear,et al.  The connectomics of brain disorders , 2015, Nature Reviews Neuroscience.

[6]  Fahad Shahbaz Khan,et al.  Self-supervised Knowledge Distillation for Few-shot Learning , 2020, BMVC.

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

[8]  Islem Rekik,et al.  Residual Embedding Similarity-Based Network Selection for Predicting Brain Network Evolution Trajectory from a Single Observation , 2020, PRIME@MICCAI.

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

[10]  Dinggang Shen,et al.  Resting-state functional MRI studies on infant brains: A decade of gap-filling efforts , 2019, NeuroImage.

[11]  Anders M. Dale,et al.  Sequence-independent segmentation of magnetic resonance images , 2004, NeuroImage.

[12]  Yi Pan,et al.  Complex Brain Network Analysis and Its Applications to Brain Disorders: A Survey , 2017, Complex..

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

[14]  Paul J. Laurienti,et al.  A New Measure of Centrality for Brain Networks , 2010, PloS one.

[15]  Ahmed Nebli,et al.  Deep EvoGraphNet Architecture For Time-Dependent Brain Graph Data Synthesis From a Single Timepoint , 2020, PRIME@MICCAI.

[16]  Dinggang Shen,et al.  Joint prediction of longitudinal development of cortical surfaces and white matter fibers from neonatal MRI , 2017, NeuroImage.

[17]  N. Tzourio-Mazoyer,et al.  Automated Anatomical Labeling of Activations in SPM Using a Macroscopic Anatomical Parcellation of the MNI MRI Single-Subject Brain , 2002, NeuroImage.

[18]  Hao Chen,et al.  Deep Semi-supervised Knowledge Distillation for Overlapping Cervical Cell Instance Segmentation , 2020, MICCAI.

[19]  Chi-Wing Fu,et al.  Difficulty-aware Meta-Learning for Rare Disease Diagnosis , 2019, MICCAI.

[20]  Edward T. Bullmore,et al.  Fundamentals of Brain Network Analysis , 2016 .

[21]  Zhiwu Lu,et al.  Large-Scale Few-Shot Learning: Knowledge Transfer With Class Hierarchy , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[22]  Leonard M. Freeman,et al.  A set of measures of centrality based upon betweenness , 1977 .

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

[24]  Gustavo Carneiro,et al.  Few-Shot Anomaly Detection for Polyp Frames from Colonoscopy , 2020, MICCAI.

[25]  Dinggang Shen,et al.  Predicting infant cortical surface development using a 4D varifold-based learning framework and local topography-based shape morphing , 2016, Medical Image Anal..

[26]  Isabelle Bloch,et al.  Knowledge Distillation from Multi-modal to Mono-modal Segmentation Networks , 2020, MICCAI.

[27]  Gang Li,et al.  Multi-Regression based supervised sample selection for predicting baby connectome evolution trajectory from neonatal timepoint , 2020, Medical Image Anal..