Deep learning for neuroimaging: a validation study

By Sergey M. Plis, R. Devon Hjelm, Ruslan Salakhutdinov, Vince D. Calhoun. 2013

604 citations recorded by the source corpus.

Deep learning methods have recently made notable advances in the tasks of classification and representation learning. These tasks are important for brain imaging and neuroscience discovery, making the methods attractive for porting to a neuroimager's toolbox. Success of these methods is, in part, explained by the flexibility of deep learning models. However, this flexibility makes the process of porting to new areas a difficult parameter optimization problem. In this work we demonstrate our results (and feasible parameter ranges) in application of deep learning methods to structural and functional brain imaging data. These methods include deep belief networks and their building block the restricted Boltzmann machine. We also describe a novel constraint-based approach to visualizing high dimensional data. We use it to analyze the effect of parameter choices on data transformations. Our results show that deep learning methods are able to learn physiologically important representations and detect latent relations in neuroimaging data.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

A survey on deep learning in medical image analysis (2017, neutral): “Plis et al. (2014) DBN Deep belief networks evaluated on brain network estimation, Schizophrenia and Huntington’s disease classification”
Interpretable Deep Neural Networks for Single-Trial EEG Classification (2016, background): “Although researchers have recently started introducing this promising technology into the domain of cognitive neuroscience [5] and Brain-Computer Interfacing (BCI) [6], [7], most of the current techniques in these fields are still based on linear methods [8], [9].”
Large-Scale Unsupervised Deep Representation Learning for Brain Structure (2018, methodology): “Plis et al. [11] presented the use of DBNs in a Constraint Satisfaction Problem framework to learn latent embeddings of gray matter images extracted from sBMRI scans on two datasets separately .”
Deep neural network with weight sparsity control and pre-training extracts hierarchical features and enhances classification performance: Evidence from whole-brain resting-state functional connectivity patterns of schizophrenia (2015, unknown): “Despite accumulating evidence showing the superiority of the DNN, previous applications of DNN to neuroimaging data are limited to only a few studies (Brosch and Tam, 2013; Hjelm et al., 2014; Plis et al., 2014; Suk et al., 2013).”
Deep Learning in Medical Image Analysis (2017, methodology): “(116) applied DBN to MR images and validated feasibility of the application by investigating if a building block of deep generative models was competitive with independent component analysis, mostly widely used method for functional MRI (fMRI) analysis.”
Analyzing Neuroimaging Data Through Recurrent Deep Learning Models (2018, unknown): “Whileresearchers have started exploring the application of DL models to neuroimaging data (e.g., [Plis et al., 2014, Mensch et al., 2018, Sarraf and Toﬁghi, 2016, Nie et al., 2016, Suk et al., 2014, Petrov et al., 2018, Yousefnezhad and Zhang, 2018]), two major challenges have so far prevented…”
EEGNet: a compact convolutional neural network for EEG-based brain–computer interfaces (2016, methodology): “ hand, rather than relying on manual feature extraction approaches. Recently, Deep Learning approaches have been used to model neurophysiological data, in functional magnetic resonance imaging (fMRI) [31] as well as EEG [32{35]. For example, Langkvist et. al. used deep belief networks (DBNs) and res”
Multimodal Neuroimaging Feature Learning With Multimodal Stacked Deep Polynomial Networks for Diagnosis of Alzheimer's Disease (2018, methodology): “also used RBM and DBN to learn features from MRI and fMRI for schizophrenia diagnosis [34].”
Deep Feature Learning for EEG Recordings (2015, background): “In this application domain, the potential of deep learning techniques for neuroimaging has been demonstrated very recently by Plis et al. (2014) for functional and structural magnetic resonance imaging (MRI) data.”
State-space model with deep learning for functional dynamics estimation in resting-state fMRI (2016, unknown): “…M anuscript A uthor M anuscript work along with recent successes in various application domains including neuroimaging analysis (Hjelm et al., 2014; Plis et al., 2014; Wu et al., 2015; Suk et al., 2014, 2015b; Kim et al., 2016), we devise a deep architecture by stacking multiple RBMs to transform…”
Preserving differential privacy in convolutional deep belief networks (2017, background): “2015a, b, 2016a, c), feature learning from fMRI data (Plis et al. 2014), diagnosis code assignment (Gottlieb et al.”
Interpretable LSTMs For Whole-Brain Neuroimaging Analyses (2018, background): “, [22, 10, 23, 24, 25, 18, 26]), three major obstacles have so far prevented broad DL usage: (1) No structured framework for the application of DL models to whole-brain neuroimaging data exists.”
Using deep learning to investigate the neuroimaging correlates of psychiatric and neurological disorders: Methods and applications (2017, unknown): “likely to be associated in complex ways, such as neuroimaging and clinical/cognitive information (Plis et al., 2014).”
Applications of Deep Learning and Reinforcement Learning to Biological Data (2017, methodology): “In addition, DBN was successfully applied to identify: Attention Deficit Hyperactivity Disorder [142], and Schizophrenia (SZ) and Huntington Disease from (f/s)MRI [122].”
Residual and plain convolutional neural networks for 3D brain MRI classification (2017, background): “Recently, there appeared several publications that use deep learning for 3D structural magnetic resonance images (MRI) processing, with particular algorithms ranging from manifold learning with Restricted Boltzmann Machines [1] to a combination of autoencoders and classifiers [2] and to meta analysi”
Deep learning in bioinformatics (2016, unknown): “[122] classified schizophrenia patients from brain MRIs using DBN, and Xu et al.”
Searching for prostate cancer by fully automated magnetic resonance imaging classification: deep learning versus non-deep learning (2017, background): “The use of AI technologies in medical imaging has becoming popular, especially with the development of deep learning 15–27 .”