Saman Sarraf

A robust and adaptive decision-making algorithm for detecting brain networks using functional MRI within the spatial and frequency domain

As the interest in functional connectivity continues to increase among neuroimaging researchers there becomes a greater need to develop an objective method of network identification. The current paper offers a solution to this problem by developing a robust decision making algorithm that can extract a target neural network from an array of spatial maps. We used a probabilistic independent component analysis to generate spatial maps of the Default Mode Network (DMN); however, this adaptive pipeline can be applied to any network of interest. Different template matching algorithms including: Normalized Cross-Correlation, Sum of Squared Differences and Dice Coefficient, were applied to the spatial and frequency domains of the dataset to identify the components that shared the greatest similarity to our DMN template. After identifying components within the resting state, the decision making pipeline selected the components within each method that had the highest matching scores to our DMN template. The final decision of selecting the most prototypical DMN components was made by a comparison between methods. This resulted in a DMN mask that was generated by the components chosen by our decision-making algorithm. To evaluate the accuracy of the decision-maker, a cross-correlation between each final mask and the template was measured. Results indicated that the Normalized Cross Correlation method, using both the spatial and frequency domain, and the Dice Coefficient method, generated the optimal DMN mask. This demonstrates the utility of our algorithm in providing an objective method for network extraction.

DeepAD: Alzheimer′s Disease Classification via Deep Convolutional Neural Networks using MRI and fMRI

To extract patterns from neuroimaging data, various statistical methods and machine learning algorithms have been explored for the diagnosis of Alzheimer’s disease among older adults in both clinical and research applications; however, distinguishing between Alzheimer’s and healthy brain data has been challenging in older adults (age > 75) due to highly similar patterns of brain atrophy and image intensities. Recently, cutting-edge deep learning technologies have rapidly expanded into numerous fields, including medical image analysis. This paper outlines state-of-the-art deep learning-based pipelines employed to distinguish Alzheimer’s magnetic resonance imaging (MRI) and functional MRI (fMRI) from normal healthy control data for a given age group. Using these pipelines, which were executed on a GPU-based high-performance computing platform, the data were strictly and carefully preprocessed. Next, scale- and shift-invariant low- to high-level features were obtained from a high volume of training images using convolutional neural network (CNN) architecture. In this study, fMRI data were used for the first time in deep learning applications for the purposes of medical image analysis and Alzheimer’s disease prediction. These proposed and implemented pipelines, which demonstrate a significant improvement in classification output over other studies, resulted in high and reproducible accuracy rates of 99.9% and 98.84% for the fMRI and MRI pipelines, respectively. Additionally, for clinical purposes, subject-level classification was performed, resulting in an average accuracy rate of 94.32% and 97.88% for the fMRI and MRI pipelines, respectively. Finally, a decision making algorithm designed for the subject-level classification improved the rate to 97.77% for fMRI and 100% for MRI pipelines.

Brain network extraction from probabilistic ICA using functional Magnetic Resonance Images and advanced template matching techniques

The human brain is a complicated network made-up of a large number of regions, which are structurally and/or functionally connected. Recently, neuroimaging studies using functional Magnetic Resonance Imaging have revealed that certain neural structures are highly active during periods of rest. Amongst several methods that have been developed to analyze resting-state fMRI data, Probabilistic Independent Component Analysis (PICA) is currently the most popular technique. The major challenge of using PICA is that resting-state networks are split into several components and visually extracting them can be difficult. In this paper, we propose a fast and precise algorithm based on advanced template matching in spatial domain such as Normalized Cross Correlation adapted to functional images in order to automatically extract the Default Mode Network (DMN) which is the task independent resting state network in the brain using PICA. We create a DMN template covering all reported regions in literature using two standard atlases. Ultimately, we reconstruct an image of the extracted DMN from PICA using an optimized decision making. Our approach was effective given that our algorithm results correlated highly with the DMN template.

Deep learning-based pipeline to recognize Alzheimer's disease using fMRI data

Over the past decade, machine learning techniques and in particular predictive modeling and pattern recognition in biomedical sciences, from drug delivery systems to medical imaging, have become one of the most important methods of assisting researchers in gaining a deeper understanding of issues in their entirety and solving complex medical problems. Deep learning is a powerful machine learning algorithm in classification that extracts low-to high-level features. In this paper, we employ a convolutional neural network to distinguish an Alzheimers brain from a normal, healthy brain. The importance of classifying this type of medical data lies in its potential to develop a predictive model or system in order to recognize the symptoms of Alzheimers disease when compared with normal subjects and to estimate the stages of the disease. Classification of clinical data for medical conditions such as Alzheimers disease has always been challenging, and the most problematic aspect has always been selecting the strongest discriminative features. Using the Convolutional Neural Network (CNN) and the famous architecture LeNet-5, we successfully classified functional MRI data of Alzheimers subjects from normal controls, where the accuracy of testing data reached 96.85%. This experiment suggests that the shift and scale invariant features extracted by CNN followed by deep learning classification represents the most powerful method of distinguishing clinical data from healthy data in fMRI. This approach also allows for expansion of the methodology to predict more complicated systems.

The associative memory deficit in aging is related to reduced selectivity of brain activity during encoding

Human aging is characterized by reductions in the ability to remember associations between items, despite intact memory for single items. Older adults also show less selectivity in task-related brain activity, such that patterns of activation become less distinct across multiple experimental tasks. This reduced selectivity or dedifferentiation has been found for episodic memory, which is often reduced in older adults, but not for semantic memory, which is maintained with age. We used fMRI to investigate whether there is a specific reduction in selectivity of brain activity during associative encoding in older adults, but not during item encoding, and whether this reduction predicts associative memory performance. Healthy young and older adults were scanned while performing an incidental encoding task for pictures of objects and houses under item or associative instructions. An old/new recognition test was administered outside the scanner. We used agnostic canonical variates analysis and split-half resampling to detect whole-brain patterns of activation that predicted item versus associative encoding for stimuli that were later correctly recognized. Older adults had poorer memory for associations than did younger adults, whereas item memory was comparable across groups. Associative encoding trials, but not item encoding trials, were predicted less successfully in older compared with young adults, indicating less distinct patterns of associative-related activity in the older group. Importantly, higher probability of predicting associative encoding trials was related to better associative memory after accounting for age and performance on a battery of neuropsychological tests. These results provide evidence that neural distinctiveness at encoding supports associative memory and that a specific reduction of selectivity in neural recruitment underlies age differences in associative memory.

Task-linked diurnal brain network reorganization in older adults: A graph theoretical approach

Testing older adults in the morning generally improves behavioral performance relative to afternoon testing. Morning testing is also associated with brain activity similar to that of young adults. Here, we used graph theory to explore how time of day (TOD) affects the organization of brain networks in older adults across rest and task states. We used nodes from the automated anatomical labeling atlas to construct participant-specific correlation matrices of fMRI data obtained during 1-back tasks with interference and rest. We computed pairwise group differences for key graph metrics, including small-worldness and modularity. We found that older adults tested in the morning and young adults did not differ on any graph metric. Both of these groups differed from older adults tested in the afternoon during the tasks—but not rest. Specifically, the latter group had lower modularity and small-worldness (indices of more efficient network organization). Across all groups, higher modularity and small-worldness strongly correlated with reduced distractibility on an implicit priming task. Increasingly, TOD is seen as important for interpreting and reproducing neuroimaging results. Our study emphasizes how TOD affects brain network organization and executive control in older adults.

Mathematical modeling for predicting betamethasone profile and burst release from in situ forming systems based on PLGA

This paper examines the application of mathematical modeling to a novel drug delivery system using artificial neural networks. For this purpose, a Feed-Forward back propagation network was trained by two different concepts and the behavior of this drug delivery system was analyzed based on the simulated results. The network also successfully determined the most accurate release profiles under specific formulation parameters. The simulated results showed a high correlation with the real data in this study. Furthermore, a new method was proposed in order to predict the burst release point in Poly Lactic-co-Glycolic Acid (PLGA) based drug delivery systems. This paper reveals that the mathematical modeling of novel drug delivery systems not only significantly decreases time and cost, but also facilitates the design of new pharmaceutical formulations.

Big data application in functional magnetic resonance imaging using apache spark

Recently, big data applications have been rapidly expanding into various industries. Healthcare is among those industries that are willing to use big data platforms, and as a result, some large data analytics tools have been adopted in this field. Medical imaging, which is a pillar in diagnostic healthcare, involves a high volume of data collection and processing. A massive number of 3D and 4D images are acquired in different forms and resolutions using a variety of medical imaging modalities. Preprocessing and analysis of imaging data is currently a costly and time-consuming process. However, few big data platforms have been created or modified for medical imaging purposes because of certain restrictions, such as data format. In this paper, we designed, developed and successfully tested a new pipeline for medical imaging data (in particular, functional magnetic resonance imaging — fMRI) using the Big Data Spark / PySpark platform on a single node, which allowed us to read and load imaging data, convert the data to Resilient Distributed Datasets in order to manipulate and perform in-memory data processing in parallel, and convert final results to an imaging format. Additionally, the pipeline provides an option to store the results in other formats, such as data frames. Using this new pipeline, we repeated our previous work, in which we extracted brain networks from fMRI data using template matching and the sum of squared differences (SSD) method. The final results revealed that our Spark (PySpark) based solution improved the performance (in terms of processing time) approximately fourfold when compared with the previous work developed in Python.