From Connectomic to Task-evoked Fingerprints: Individualized Prediction\n of Task Contrasts from Resting-state Functional Connectivity

2020/08/06 by Gia H. Ngo, Ngo, Gia H., Meenakshi Khosla +7 · 1 citation
Medicine · Neuroscience · Physics and Astronomy · #Advanced MRI Techniques and Applications #Atomic and Subatomic Physics Research #FOS: Biological sciences #FOS: Computer and information sciences #Functional Brain Connectivity Studies #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Neurons and Cognition (q-bio.NC)

paper · pdf · doi:10.48550/arxiv.2008.02961

openalex publication_date 2020/08/06 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28

Abstract

Resting-state functional MRI (rsfMRI) yields functional connectomes that can\nserve as cognitive fingerprints of individuals. Connectomic fingerprints have\nproven useful in many machine learning tasks, such as predicting\nsubject-specific behavioral traits or task-evoked activity. In this work, we\npropose a surface-based convolutional neural network (BrainSurfCNN) model to\npredict individual task contrasts from their resting-state fingerprints. We\nintroduce a reconstructive-contrastive loss that enforces subject-specificity\nof model outputs while minimizing predictive error. The proposed approach\nsignificantly improves the accuracy of predicted contrasts over a\nwell-established baseline. Furthermore, BrainSurfCNN's prediction also\nsurpasses test-retest benchmark in a subject identification task.\n

Cited by

• Rest2Visual: Predicting Visually Evoked fMRI from Resting-State Scans

Related

• Functional connectome fingerprinting: Identifying individuals and predicting cognitive function via deep learning
• High-Accuracy Machine Learning Techniques for Functional Connectome Fingerprinting and Cognitive State Decoding
• Constructing Compact Brain Connectomes for Individual Fingerprinting
• Zero-shot Learning of Individualized Task Contrast Prediction from Resting-state Functional Connectomes
• Functional Connectome Fingerprinting Using Convolutional and Dictionary Learning