Deep neural network with weight sparsity control and pre-training extracts hierarchical features and enhances classification performance: Evidence from whole-brain resting-state functional…

By Junghoe Kim, Vince D. Calhoun, Eunsoo Shim, Jong-Hwan Lee. NeuroImage, 2015

330 citations recorded by the source corpus.

Functional connectivity (FC) patterns obtained from resting-state functional magnetic resonance imaging data are commonly employed to study neuropsychiatric conditions by using pattern classifiers such as the support vector machine (SVM). Meanwhile, a deep neural network (DNN) with multiple hidden layers has shown its ability to systematically extract lower-to-higher level information of image and speech data from lower-to-higher hidden layers, markedly enhancing classification accuracy. The objective of this study was to adopt the DNN for whole-brain resting-state FC pattern classification of schizophrenia (SZ) patients vs. healthy controls (HCs) and identification of aberrant FC patterns associated with SZ. We hypothesized that the lower-to-higher level features learned via the DNN would significantly enhance the classification accuracy, and proposed an adaptive learning algorithm to explicitly control the weight sparsity in each hidden layer via L1-norm regularization. Furthermore, the weights were initialized via stacked autoencoder based pre-training to further improve the classification performance. Classification accuracy was systematically evaluated as a function of (1) the number of hidden layers/nodes, (2) the use of L1-norm regularization, (3) the use of the pre-training, (4) the use of framewise displacement (FD) removal, and (5) the use of anatomical/functional parcellation. Using FC patterns from anatomically parcellated regions without FD removal, an error rate of 14.2% was achieved by employing three hidden layers and 50 hidden nodes with both L1-norm regularization and pre-training, which was substantially lower than the error rate from the SVM (22.3%). Moreover, the trained DNN weights (i.e., the learned features) were found to represent the hierarchical organization of aberrant FC patterns in SZ compared with HC. Specifically, pairs of nodes extracted from the lower hidden layer represented sparse FC patterns implicated in SZ, which was quantified by using kurtosis/modularity measures and features from the higher hidden layer showed holistic/global FC patterns differentiating SZ from HC. Our proposed schemes and reported findings attained by using the DNN classifier and whole-brain FC data suggest that such approaches show improved ability to learn hidden patterns in brain imaging data, which may be useful for developing diagnostic tools for SZ and other neuropsychiatric disorders and identifying associated aberrant FC patterns.

Subfields: Cognitive Neuroscience, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/996e3aad75caff4926f54ec6a3753df4a3a60981

How researchers see this paper

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

In Cognitive Neuroscience

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

A survey on deep learning in medical image analysis (2017, methodology): “The two papers that do, Antony et al. (2016) and Kim et al. (2016a), o ﬀ er con-ﬂicting results.”
A survey of deep neural network architectures and their applications (2017, methodology): “Recently, the DNNs have been employed to classify the whole-brain resting-state FC patterns of schizophrenia (SZ) [83].”
Single Subject Prediction of Brain Disorders in Neuroimaging: Promises and Pitfalls (2016, background): “They showed that their approach outperforms SVM by a significant margin (Kim et al., 2015).”
Using deep learning to investigate the neuroimaging correlates of psychiatric and neurological disorders: Methods and applications (2017, unknown): “neuroimaging studies are using DL to elucidate the neural correlates of these disorders (e.g. Payan and Montana, 2015; Pills et al., 2014; Kim et al., 2016).”
Machine Learning in Major Depression: from Classification to Treatment Outcome Prediction. (2018, methodology): “For example, Kim et al adopt deep neural network (DNN) with L1‐norm controlling weight sparsity in hidden layer for whole‐brain resting‐state FC pat‐ tern classification of schizophrenia vs HC. 116 Zeng et al investigated cross‐site classification with deep learning in schizophrenia for the first ti”
Classification and Prediction of Brain Disorders Using Functional Connectivity: Promising but Challenging (2018, methodology): “Deep Learning Classifiers Deep learning methods have attracted increasing interesting in various areas and also have been applied in the classification of brain disorders (Plis et al., 2014; Iidaka, 2015; Lecun et al., 2015; Calhoun and Sui, 2016; Hu et al., 2016; Kim et al., 2016; Han et al., 2017;”
A Novel Transfer Learning Approach to Enhance Deep Neural Network Classification of Brain Functional Connectomes (2018, methodology): “…such as deep neural network (DNN) (Hinton and Salakhutdinov, 2006; Hinton et al., 2006), on rs-fMRI data to extract high-level FC features for the classiﬁcation of neuropsychiatric and neurodevelopmental disorders (Kuang et al., 2014; Kim et al., 2016; Guo et al., 2017; Heinsfeld et al., 2018).”
Multi-Site Diagnostic Classification of Schizophrenia Using Discriminant Deep Learning with Functional Connectivity MRI (2018, methodology): “The training dataset was further randomly partitioned into nine folds, and eight folds were used for training and one fold for validation in the DANS and SAN classi ﬁ cation frameworks.”
State-space model with deep learning for functional dynamics estimation in resting-state fMRI (2016, unknown): “Justified by their 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 the input featu”