A robust variational autoencoder using beta divergence.

The presence of outliers can severely degrade learned representations and performance of deep learning methods and hence disproportionately affect the training process, leading to incorrect conclusions about the data. For example, anomaly detection using deep generative models is typically only possible when similar anomalies (or outliers) are not present in the training data. Here we focus on variational autoencoders (VAEs). While the VAE is a popular framework for anomaly detection tasks, we observe that the VAE is unable to detect outliers when the training data contains anomalies that have the same distribution as those in test data. In this paper we focus on robustness to outliers in training data in VAE settings using concepts from robust statistics. We propose a variational lower bound that leads to a robust VAE model that has the same computational complexity as the standard VAE and contains a single automatically-adjusted tuning parameter to control the degree of robustness. We present mathematical formulations for robust variational autoencoders (RVAEs) for Bernoulli, Gaussian and categorical variables. The RVAE model is based on beta-divergence rather than the standard Kullback-Leibler (KL) divergence. We demonstrate the performance of our proposed ß-divergence-based autoencoder for a variety of image and categorical datasets showing improved robustness to outliers both qualitatively and quantitatively. We also illustrate the use of our robust VAE for detection of lesions in brain images, formulated as an anomaly detection task. Finally, we suggest a method to tune the hyperparameter of RVAE which makes our model completely unsupervised.

Meta Transfer of Self-Supervised Knowledge: Foundation Model in Action for Post-Traumatic Epilepsy Prediction.

Despite the impressive advancements achieved using deep-learning for functional brain activity analysis, the heterogeneity of functional patterns and scarcity of imaging data still pose challenges in tasks such as prediction of future onset of Post-Traumatic Epilepsy (PTE) from data acquired shortly after traumatic brain injury (TBI). Foundation models pre-trained on separate large-scale datasets can improve the performance from scarce and heterogeneous datasets. For functional Magnetic Resonance Imaging (fMRI), while data may be abundantly available from healthy controls, clinical data is often scarce, limiting the ability of foundation models to identify clinically-relevant features. We overcome this limitation by introducing a novel training strategy for our foundation model by integrating meta-learning with self-supervised learning to improve the generalization from normal to clinical features. In this way we enable generalization to other downstream clinical tasks, in our case prediction of PTE. To achieve this, we perform self-supervised training on the control dataset to focus on inherent features that are not limited to a particular supervised task while applying meta-learning, which strongly improves the model's generalizability using bi-level optimization. Through experiments on neurological disorder classification tasks, we demonstrate that the proposed strategy significantly improves task performance on small-scale clinical datasets. To explore the generalizability of the foundation model in downstream applications, we then apply the model to an unseen TBI dataset for prediction of PTE using zero-shot learning. Results further demonstrated the enhanced generalizability of our foundation model.

Deep Quantile Regression for Uncertainty Estimation in Unsupervised and Supervised Lesion Detection.

Despite impressive state-of-the-art performance on a wide variety of machine learning tasks in multiple applications, deep learning methods can produce over-confident predictions, particularly with limited training data. Therefore, quantifying uncertainty is particularly important in critical applications such as lesion detection and clinical diagnosis, where a realistic assessment of uncertainty is essential in determining surgical margins, disease status and appropriate treatment. In this work, we propose a novel approach that uses quantile regression for quantifying aleatoric uncertainty in both supervised and unsupervised lesion detection problems. The resulting confidence intervals can be used for lesion detection and segmentation. In the unsupervised setting, we combine quantile regression with the Variational AutoEncoder (VAE). The VAE is trained on lesion-free data, so when presented with an image with a lesion, it tends to reconstruct a lesion-free version of the image. To detect the lesion, we then compare the input (lesion) and output (lesion-free) images. Here we address the problem of quantifying uncertainty in the images that are reconstructed by the VAE as the basis for principled outlier or lesion detection. The VAE models the output as a conditionally independent Gaussian characterized by its mean and variance. Unfortunately, joint optimization of both mean and variance in the VAE leads to the well-known problem of shrinkage or underestimation of variance. Here we describe an alternative Quantile-Regression VAE (QR-VAE) that avoids this variance shrinkage problem by directly estimating conditional quantiles for the input image. Using the estimated quantiles, we compute the conditional mean and variance for the input image from which we then detect outliers by thresholding at a false-discovery-rate corrected p-value. In the supervised setting, we develop binary quantile regression (BQR) for the supervised lesion segmentation task. We show how BQR can be used to capture uncertainty in lesion boundaries in a manner that characterizes expert disagreement.

Differences in behavioral and cortical indices in pianists and non-musicians during a non-musical motor planning task: An event-related potential study.

Psychological studies have shown that music has an impact on human cognitive function. We aimed to compare the performance and neural activity of pianists and non-musicians during a non-musical motor-planning task. In addition, we investigated the effect of task complexity on the characteristics of the behavioral and neural responses. The participants had to grasp a hexagonal knob with their right hand and rotate it 60° or 180° clockwise (CW) or counterclockwise (CCW). We examined the groups in terms of the amplitude of the P2 component in the event-related potential (at the neural level) and the planning time, grasping time, releasing time, and planning pattern for initial grip selection (at the behavioral level). At the behavioral level, we observed no significant difference between groups, while at the neural level; we found an interaction between direction and group indicating that pianists showed lower P2 amplitude in the CW directions. However, there was no significant difference between groups in the CCW direction. A significant main effect of rotation was revealed at both the neural and behavioral levels; increasing the rotation angle led to an increase in the planning time and the P2 amplitude, indicating a complexity effect. In conclusion, we observed that pianists had lower P2 amplitude in lateral movements than non-musicians; however, due to the lack of behavioral group differences, further research is warranted to support the far-transfer theory in this field.

Quantile Regression for Uncertainty Estimation in VAEs with Applications to Brain Lesion Detection.

The Variational AutoEncoder (VAE) has become one of the most popular models for anomaly detection in applications such as lesion detection in medical images. The VAE is a generative graphical model that is used to learn the data distribution from samples and then generate new samples from this distribution. By training on normal samples, the VAE can be used to detect inputs that deviate from this learned distribution. The VAE models the output as a conditionally independent Gaussian characterized by means and variances for each output dimension. VAEs can therefore use reconstruction probability instead of reconstruction error for anomaly detection. Unfortunately, joint optimization of both mean and variance in the VAE leads to the well-known problem of shrinkage or underestimation of variance. We describe an alternative VAE model, Quantile-Regression VAE (QR-VAE), that avoids this variance shrinkage problem by estimating conditional quantiles for the given input image. Using the estimated quantiles, we compute the conditional mean and variance for input images under the Gaussian model. We then compute reconstruction probability using this model as a principled approach to outlier or anomaly detection. We also show how our approach can be used for heterogeneous thresholding of images for detecting lesions in brain images.

BRAIN LESION DETECTION USING A ROBUST VARIATIONAL AUTOENCODER AND TRANSFER LEARNING.

Automated brain lesion detection from multi-spectral MR images can assist clinicians by improving sensitivity as well as specificity. Supervised machine learning methods have been successful in lesion detection. However, these methods usually rely on a large number of manually delineated images for specific imaging protocols and parameters and often do not generalize well to other imaging parameters and demographics. Most recently, unsupervised models such as autoencoders have become attractive for lesion detection since they do not need access to manually delineated lesions. Despite the success of unsupervised models, using pre-trained models on an unseen dataset is still a challenge. This difficulty is because the new dataset may use different imaging parameters, demographics, and different pre-processing techniques. Additionally, using a clinical dataset that has anomalies and outliers can make unsupervised learning challenging since the outliers can unduly affect the performance of the learned models. These two difficulties make unsupervised lesion detection a particularly challenging task. The method proposed in this work addresses these issues using a two-prong strategy: (1) we use a robust variational autoencoder model that is based on robust statistics, specifically the ß-divergence that can be trained with data that has outliers; (2) we use a transfer-learning method for learning models across datasets with different characteristics. Our results on MRI datasets demonstrate that we can improve the accuracy of lesion detection by adapting robust statistical models and transfer learning for a variational autoencoder model.

Predicting Cognitive Scores from Resting fMRI Data and Geometric Features of the Brain.

Anatomical T1 weighted Magnetic Resonance Imaging (MRI) and functional magnetic resonance imaging collected during resting (rfMRI) are promising markers that offer insight into the structure and function of the human brain. The objective of this work is to explore the use of a deep learning neural network to predict cognitive performance scores for a population of normal controls and subjects with Attention Deficit Hyperactivity Disorder (ADHD). Specifically, we predict verbal and performance IQs and ADHD index from features derived from T1 and rfMRI imaging data. First, we processed the rfMRI and MRI data of subjects using the BrainSuite fMRI Processing (BFP) pipeline to perform anatomical and functional preprocessing. This produces for each subject fMRI and geometric (anatomical) features represented in a standardized grayordinate system. The geometric and functional cortical data corresponding to the two hemispheres were then transformed to 128×128 multichannel images and input to a convolutional component of the neural network. Subcortical data were presented in a standard vector form and inputted to a input layer of the network. The neural network was implemented in Python using the Keras library with a TensorFlow backend. Training was performed on 168 images with 90 images used for testing. We observed a high correlation between predicted and actual values of the indices tested: Performance IQ: 0.47; Verbal IQ: 0.41, ADHD: 0.57. Comparing these values to those from network trained on functional-only and structural-only data, we saw that rfMRI is more informative than MRI, but the two modalities are highly complementary in terms of predicting these indices.

A Matched Filter Decomposition of fMRI into Resting and Task Components.

The human brain exhibits dynamic interactions among brain regions when responding to stimuli and executing tasks, which can be recorded using functional magnetic resonance imaging (fMRI). Functional MRI signals collected in response to specific tasks consist of a combination of task-related and spontaneous (task-independent) activity. By exploiting the highly structured spatiotemporal patterns of resting state networks, this paper presents a matched-filter approach to decomposing fMRI signals into task and resting-state components. To perform the decomposition, we first use a temporal alignment procedure that is a windowed version of the brainsync transform to synchronize a resting template to the brain's response to tasks. The resulting 'matched filter' removes the components of the fMRI signal that can be described by resting connectivity, leaving the portion of brain activity directly related to tasks. We present a closed-form expression for the windowed synchronization transform that is used by the matched filter. We demonstrate performance of this procedure in application to motor task and language task fMRI data. We show qualitatively and quantitatively that by removing the resting activity, we are able to identify task activated regions in the brain more clearly. Additionally, we show improved prediction accuracy in multivariate pattern analysis when using the matched filtered fMRI data.

Lost in music: Neural signature of pleasure and its role in modulating attentional resources.

We investigated the neural correlates of pleasure induced by listening to highly pleasant and neutral musical excerpts using electroencephalography (EEG). Power spectrum analysis of EEG data showed a distinct gradual change in the power of low-frequency oscillations in response to highly pleasant, but not neutral, musical excerpts. Specifically, listening to highly pleasant music was associated with (i) relatively higher oscillatory activity in the theta band over the frontocentral (FC) area and in the alpha band over the parieto-occipital area, and (ii) a gradual increase in the oscillatory power over time. Correlation analysis between behavioral and electrophysiological data revealed that theta power over the FC electrodes was correlated with subjective assessment of pleasantness while listening to music. To study the link between attention and positive valence in our experiments, volunteers performed a delayed match-to-sample memory task while listening to the musical excerpts. The subjects' performances were significantly lower under highly pleasant conditions compared to neutral conditions. Listening to pleasant music requires higher degrees of attention, leading to the observed decline in memory performance. Gradual development of low-frequency oscillations in the frontal and posterior areas may be at least partly due to gradual recruitment of higher levels of attention over time in response to pleasurable music.

rfDemons: Resting fMRI-based Cortical Surface Registration using the BrainSync Transform.

Cross subject functional studies of cerebral cortex require cortical registration that aligns functional brain regions. While cortical folding patterns are approximate indicators of the underlying cytoarchitecture, coregistration based on these features alone does not accurately align functional regions in cerebral cortex. This paper presents a method for cortical surface registration (rfDemons) based on resting fMRI (rfMRI) data that uses curvature-based anatomical registration as an initialization. In contrast to existing techniques that use connectivity-based features derived from rfMRI, the proposed method uses 'synchronized' resting rfMRI time series directly. The synchronization of rfMRI data is performed using the BrainSync transform which applies an orthogonal transform to the rfMRI time series to temporally align them across subjects. The rfDemons method was applied to rfMRI from the Human Connectome Project and evaluated using task fMRI data to explore the impact of cortical registration performed using resting fMRI data on functional alignment of the cerebral cortex.

Culture Modulates the Brain Response to Harmonic Violations: An EEG Study on Hierarchical Syntactic Structure in Music.

We investigated the role of culture in processing hierarchical syntactic structures in music. We examined whether violation of non-local dependencies manifest in event related potentials (ERP) for Western and Iranian excerpts by recording EEG while participants passively listened to sequences of modified/original excerpts. We also investigated oscillatory and synchronization properties of brain responses during processing of hierarchical structures. For the Western excerpt, subjective ratings of conclusiveness were marginally significant and the difference in the ERP components fell short of significance. However, ERP and behavioral results showed that while listening to culturally familiar music, subjects comprehended whether or not the hierarchical syntactic structure was fulfilled. Irregularities in the hierarchical structures of the Iranian excerpt elicited an early negativity in the central regions bilaterally, followed by two later negativities from 450-700 to 750-950 ms. The latter manifested throughout the scalp. Moreover, violations of hierarchical structure in the Iranian excerpt were associated with (i) an early decrease in the long range alpha phase synchronization, (ii) an early increase in the oscillatory activity in the beta band over the central areas, and (iii) a late decrease in the theta band phase synchrony between left anterior and right posterior regions. Results suggest that rhythmic structures and melodic fragments, representative of Iranian music, created a familiar context in which recognition of complex non-local syntactic structures was feasible for Iranian listeners. Processing of neural responses to the Iranian excerpt indicated neural mechanisms for processing of hierarchical syntactic structures in music at different levels of cortical integration.