Deep neural networks and kernel regression achieve comparable accuracies for functional connectivity prediction of behavior and demographics

• Published in: NeuroImage (Orlando, Fla.) Vol. 206; p. 116276
• Main Authors: He, Tong, Kong, Ru, Holmes, Avram J., Nguyen, Minh, Sabuncu, Mert R., Eickhoff, Simon B., Bzdok, Danilo, Feng, Jiashi, Yeo, B.T. Thomas
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.02.2020; Elsevier Limited; Elsevier
• Subjects: Adult; Age; Age Factors; Aged; Algorithms; Artificial Intelligence; Biobanks; Biological Specimen Banks; Brain; Brain - diagnostic imaging; Brain - physiology; Computational neuroscience; Computer Science; Connectome - methods; Datasets; Datasets as Topic; Deep Learning; Demography; Female; Fingerprinting; Graph convolutional neural network; Humans; Image Interpretation, Computer-Assisted - methods; Intelligence; Intelligence - physiology; Kernel ridge regression; Learning algorithms; Machine Learning; Magnetic Resonance Imaging - methods; Male; Medical imaging; Middle Aged; Neural networks; Neural Networks, Computer; Neuroimaging; Neurosciences; Phenotypes; Psychomotor Performance - physiology; Resting-state fMRI; Sex Factors; Young Adult; United Kingdom > UK; Deep learning; Resting-state fMRI; Kernel ridge regression; Fingerprinting; Graph convolutional neural network
• ISSN: 1053-8119; 1095-9572; 1095-9572
• DOI: 10.1016/j.neuroimage.2019.116276

There is significant interest in the development and application of deep neural networks (DNNs) to neuroimaging data. A growing literature suggests that DNNs outperform their classical counterparts in a variety of neuroimaging applications, yet there are few direct comparisons of relative utility. Here, we compared the performance of three DNN architectures and a classical machine learning algorithm (kernel regression) in predicting individual phenotypes from whole-brain resting-state functional connectivity (RSFC) patterns. One of the DNNs was a generic fully-connected feedforward neural network, while the other two DNNs were recently published approaches specifically designed to exploit the structure of connectome data. By using a combined sample of almost 10,000 participants from the Human Connectome Project (HCP) and UK Biobank, we showed that the three DNNs and kernel regression achieved similar performance across a wide range of behavioral and demographic measures. Furthermore, the generic feedforward neural network exhibited similar performance to the two state-of-the-art connectome-specific DNNs. When predicting fluid intelligence in the UK Biobank, performance of all algorithms dramatically improved when sample size increased from 100 to 1000 subjects. Improvement was smaller, but still significant, when sample size increased from 1000 to 5000 subjects. Importantly, kernel regression was competitive across all sample sizes. Overall, our study suggests that kernel regression is as effective as DNNs for RSFC-based behavioral prediction, while incurring significantly lower computational costs. Therefore, kernel regression might serve as a useful baseline algorithm for future studies.