Milad Makkie

Scalable Fast Rank-1 Dictionary Learning for fMRI Big Data Analysis

It has been shown from various functional neuroimaging studies that sparsity-regularized dictionary learning could achieve superior performance in decomposing comprehensive and neuroscientifically meaningful functional networks from massive fMRI signals. However, the computational cost for solving the dictionary learning problem has been known to be very demanding, especially when dealing with large-scale data sets. Thus in this work, we propose a novel distributed rank-1 dictionary learning (D-r1DL) model and apply it for fMRI big data analysis. The model estimates one rank-1 basis vector with sparsity constraint on its loading coefficient from the input data at each learning step through alternating least squares updates. By iteratively learning the rank-1 basis and deflating the input data at each step, the model is then capable of decomposing the whole set of functional networks. We implement and parallelize the rank-1 dictionary learning algorithm using Spark engine and deployed the resilient distributed dataset (RDDs) abstracts for the data distribution and operations. Experimental results from applying the model on the Human Connectome Project (HCP) data show that the proposed D-r1DL model is efficient and scalable towards fMRI big data analytics, thus enabling data-driven neuroscientific discovery from massive fMRI big data in the future.

HAFNI-enabled largescale platform for neuroimaging informatics (HELPNI).

Tremendous efforts have thus been devoted on the establishment of functional MRI informatics systems that recruit a comprehensive collection of statistical/computational approaches for fMRI data analysis. However, the state-of-the-art fMRI informatics systems are especially designed for specific fMRI sessions or studies of which the data size is not really big, and thus has difficulty in handling fMRI ‘big data.’ Given the size of fMRI data are growing explosively recently due to the advancement of neuroimaging technologies, an effective and efficient fMRI informatics system which can process and analyze fMRI big data is much needed. To address this challenge, in this work, we introduce our newly developed informatics platform, namely, ‘HAFNI-enabled largescale platform for neuroimaging informatics (HELPNI).’ HELPNI implements our recently developed computational framework of sparse representation of whole-brain fMRI signals which is called holistic atlases of functional networks and interactions (HAFNI) for fMRI data analysis. HELPNI provides integrated solutions to archive and process large-scale fMRI data automatically and structurally, to extract and visualize meaningful results information from raw fMRI data, and to share open-access processed and raw data with other collaborators through web. We tested the proposed HELPNI platform using publicly available 1000 Functional Connectomes dataset including over 1200 subjects. We identified consistent and meaningful functional brain networks across individuals and populations based on resting state fMRI (rsfMRI) big data. Using efficient sampling module, the experimental results demonstrate that our HELPNI system has superior performance than other systems for large-scale fMRI data in terms of processing and storing the data and associated results much faster.

Modeling Task fMRI Data Via Deep Convolutional Autoencoder

Task-based fMRI (tfMRI) has been widely used to study functional brain networks. Modeling tfMRI data is challenging due to at least two problems: the lack of the ground truth of underlying neural activity and the intrinsic structure of tfMRI data is highly complex. To better understand brain networks based on fMRI data, data-driven approaches were proposed, for instance, Independent Component Analysis (ICA) and Sparse Dictionary Learning (SDL). However, both ICA and SDL only build shallow models, and they are under the strong assumption that original fMRI signal could be linearly decomposed into time series components with their corresponding spatial maps. As growing evidence shows that human brain function is hierarchically organized, new approaches that can infer and model the hierarchical structure of brain networks are widely called for. Recently, deep convolutional neural network (CNN) has drawn much attention, in that deep CNN has been proven to be a powerful method for learning high-level and mid-level abstractions from low-level raw data. Inspired by the power of deep CNN, in this study, we developed a new neural network structure based on CNN, called Deep Convolutional Auto-Encoder (DCAE), in order to take the advantages of both data-driven approach and CNN’s hierarchical feature abstraction ability for the purpose of learning mid-level and high-level features from complex tfMRI time series in an unsupervised manner. The DCAE has been applied and tested on the publicly available human connectome project (HCP) tfMRI datasets, and promising results are achieved.

Signal sampling for efficient sparse representation of resting state FMRI data.

As the size of brain imaging data such as fMRI grows explosively, it provides us with unprecedented and abundant information about the brain. How to reduce the size of fMRI data but not lose much information becomes a more and more pressing issue. Recent literature studies tried to deal with it by dictionary learning and sparse representation methods, however, their computation complexities are still high, which hampers the wider application of sparse representation method to large scale fMRI datasets. To effectively address this problem, this work proposes to represent resting state fMRI (rs-fMRI) signals of a whole brain via a statistical sampling based sparse representation. First we sampled the whole brain’s signals via different sampling methods, then the sampled signals were aggregate into an input data matrix to learn a dictionary, finally this dictionary was used to sparsely represent the whole brain’s signals and identify the resting state networks. Comparative experiments demonstrate that the proposed signal sampling framework can speed-up by ten times in reconstructing concurrent brain networks without losing much information. The experiments on the 1000 Functional Connectomes Project further demonstrate its effectiveness and superiority.

Modeling 4D fMRI Data via Spatio-Temporal Convolutional Neural Networks (ST-CNN)

Simultaneous modeling of the spatio-temporal variation patterns of brain functional network from 4D fMRI data has been an important yet challenging problem for the field of cognitive neuroscience and medical image analysis. Inspired by the recent success in applying deep learning for functional brain decoding and encoding, in this work we propose a spatio-temporal convolutional neural network (ST-CNN)to jointly learn the spatial and temporal patterns of targeted network from the training data and perform automatic, pin-pointing functional network identification. The proposed ST-CNN is evaluated by the task of identifying the Default Mode Network (DMN) from fMRI data. Results show that while the framework is only trained on one fMRI dataset,it has the sufficient generalizability to identify the DMN from different populations of data as well as different cognitive tasks. Further investigation into the results show that the superior performance of ST-CNN is driven by the jointly-learning scheme, which capture the intrinsic relationship between the spatial and temporal characteristic of DMN and ensures the accurate identification.

Fast and Scalable Distributed Deep Convolutional Autoencoder for fMRI Big Data Analytics

In recent years, analyzing task-based fMRI (tfMRI) data has become an essential tool for understanding brain function and networks. However, due to the sheer size of tfMRI data, its intrinsic complex structure, and lack of ground truth of underlying neural activities, modeling tfMRI data is hard and challenging. Previously proposed data modeling methods including Independent Component Analysis (ICA) and Sparse Dictionary Learning only provided shallow models based on blind source separation under the strong assumption that original fMRI signals could be linearly decomposed into time series components with corresponding spatial maps. Given the Convolutional Neural Network (CNN) successes in learning hierarchical abstractions from low-level data such as tfMRI time series, in this work we propose a novel scalable distributed deep CNN autoencoder model and apply it for fMRI big data analysis. This model aims to both learn the complex hierarchical structures of the tfMRI big data and to leverage the processing power of multiple GPUs in a distributed fashion. To deploy such a model, we have created an enhanced processing pipeline on the top of Apache Spark and Tensorflow, leveraging from a large cluster of GPU nodes over cloud. Experimental results from applying the model on the Human Connectome Project (HCP) data show that the proposed model is efficient and scalable toward tfMRI big data modeling and analytics, thus enabling data-driven extraction of hierarchical neuroscientific information from massive fMRI big data.

Template-guided Functional Network Identification via Supervised Dictionary Learning

Functional network analysis based on matrix decomposition/factorization methods including ICA and dictionary learning models have become a popular approach in fMRI study. Yet it is still a challenging issue in interpreting the result networks because of the inter-subject variability and image noises, thus in many cases, manual inspection on the obtained networks is needed. Aiming to provide a fast and reliable functional network identification tool for both normal and diseased brain fMRI data analysis, in this work, we propose a novel supervised dictionary learning model based on rank-1 matrix decomposition algorithm (S-r1DL) with sparseness constraint. Application on the Autism Brain Imaging Data Exchange (ABIDE) database showed that S-r1DL can fast and accurately identify the functional networks based on the given templates, comparing to unsupervised learning method.

Distributed rank-1 dictionary learning: Towards fast and scalable solutions for fMRI big data analytics

The use of functional brain imaging for research and diagnosis has benefitted greatly from the recent advancements in neuroimaging technologies, as well as the explosive growth in size and availability of fMRI data. While it has been shown in literature that using multiple and large scale fMRI datasets can improve reproducibility and lead to new discoveries, the computational and informatics systems supporting the analysis and visualization of such fMRI big data are extremely limited and largely under-discussed. We propose to address these shortcomings in this work, based on previous success in using dictionary learning method for functional network decomposition studies on fMRI data. We presented a distributed dictionary learning framework based on rank-1 matrix decomposition with sparseness constraint (D-r1DL framework). The framework was implemented using the Spark distributed computing engine and deployed on three different processing units: an in-house server, in-house high performance clusters, and the Amazon Elastic Compute Cloud (EC2) service. The whole analysis pipeline was integrated with our neuroinformatics system for data management, user input/output, and real-time visualization. Performance and accuracy of D-r1DL on both individual and group-wise fMRI Human Connectome Project (HCP) dataset shows that the proposed framework is highly scalable. The resulting group-wise functional network decompositions are highly accurate, and the fast processing time confirm this claim. In addition, D-r1DL can provide real-time user feedback and results visualization which are vital for large-scale data analysis.

Implementing dictionary learning in Apache Flink, Or: How I learned to relax and love iterations

The authors evaluate the use of Apache Flink, a novel data analysis framework offering optimizations over competitors such as Apache Spark, in order to use a rank-1 dictionary learning (r1DL) algorithm to decompose fMRI data. We first expand the functionality of the Flink Python API in order to accommodate the implementation of rank-1 dictionary learning, a model for decomposing a large matrix. Iterative algorithms, aggregators, and other features are added to the incomplete Python API, and the experiences and lessons learned are described. Using these features, we port an existing implementation of r1DL from using the Python API of Apache Spark to using the Python API of Apache Flink. In preliminary testing, this implementation suggests performance boosts over Spark for large input files, meriting further research. We conclude that Flink is likely a feasible tool for the application of dictionary learning to decompose fMRI data, and we continue to evaluate and apply it.