Currently, the quantification of directed information transfer between interacting brain areas is one of the most challenging methodological problems in computational neuroscience. A fundamental problem is identifying connectivity in very high-dimensional systems. The common practice has been to transform a high-dimensional system into a simplified representation, e.g. by clustering, Principal, or Independent Component Analysis. The drawback of such methodology is that an identified interaction between components cannot readily be transferred into the original high dimensional space. Thus, directed interactions between the original network nodes can no longer be revealed. Although this significantly limits the interpretation of brain network activities in physiological and diseae states, surprisingly little effort has been devoted to circumvent the inevitable information loss induced by the aforementioned currently employed techniques. Here, we propose a novel computational framework to explore the directed information transfer in the human brain (Aim 1). In contrast to existing approaches that attempt to analyze complex systems by simplification and subsequent analysis of a simplified system, we propose to retain the original system complexity during the analysis, thus preserving essential information that is lost in initial simplification steps of currently used techniques. To this end, we introduce a large scale Granger Causality (lsGC) approach that permits connectivity analysis in large time-series ensembles obtained from functional MRI neuro-imaging studies. By avoiding irreversible information loss and retaining the original system complexity, our approach permits the detection of directed interactions between the network nodes in the original high-dimensional system. The resulting huge Granger causality matrices, however, prohibit a straightforward interpretation of the connectivity pattern within the underlying complex system. This is attributed to their size which currently exceeds human information processing capabilities. To address this problem, we propose to perform subsequent data reduction by advanced partitioning and dimensionality reduction techniques, in order to unveil information hidden in such high dimensional matrices, and to convert such information into useful real-world knowledge (Aim 2). In contrast to existing approaches, we do not use data reduction to simplify the network prior to its analysis, but only to visualize the network analysis results, after the full functional connectivity structure has been unveiled. This will allow us to create a visual mapping of the whole-brain functional connectivity network at a pixel-scale resolution. Besides facilitating neurophysiologic insight into cerebral information processing dynamics, such a comprehensive whole-brain network representation is expected to provide quantitative clinical biomarkers for evaluating disease progression and therapeutic effectiveness of clinical interventions, which may be behavioral or pharmacological. To demonstrate the applicability of our lsGC network analysis approach in a clinically relevant context, we propose a pilot study to detect changes in brain network connectivity patterns associated with starting antiretroviral therapy in antiretrovirl na?ve individuals with HIV associated cognitive impairment (Aim 3). Here, HIV-associated brain injury is not only chosen for its paramount clinical relevance (over 33 million people infected wit specific relevance in women and underrepresented minorities), but also for the existence of objectively quantifiable observables in this clinical setting, such as HIV plasma and CSF viral load, and objective neurological scores for clinically measuring cognitive impairment. The proposed pilot study will establish a clinical model for evaluating our lsGC whole-brain functional connectivity analysis as a novel biomarker for disease progression and therapy effectiveness. Intellectual merit: Based on our previous work and the outstanding infrastructural resources at both hosting institutions, our multi-disciplinary, multi-national, and multi-lingual collaboration team offers a unique combination of skills and expertise to accomplish the specific aims of this proposal. Once established, our framework will be applicable to a wide scope of neurological and psychiatric disorders as well. Beyond its immediate applications in computational neuroscience, neurophysiology, and clinical neurology, we conjecture that our approach to analyzing effective connectivity in complex systems will be useful in many other research domains throughout science and engineering, ranging from information retrieval to systems biology. Broader impacts: Our project will create a transatlantic network promoting teaching, training, and career mentoring activities for young researchers at undergraduate, graduate, and postdoctoral levels. Besides encouraging the participation of women and underrepresented minorities as research personnel, we have selected the clinical focus of the pilot study in Aim 3 because of its specific relevance to these population groups. We will publicly disseminate scientific results and research infrastructure resources, which will ensure the translational research effectiveness of our project for sustainable clinical and educational use.

Agency
National Institute of Health (NIH)
Institute
National Institute on Drug Abuse (NIDA)
Type
Research Project (R01)
Project #
5R01DA034977-02
Application #
8514556
Study Section
Special Emphasis Panel (ZRG1-IFCN-B (55))
Program Officer
Lin, Yu
Project Start
2012-08-01
Project End
2015-06-30
Budget Start
2013-07-01
Budget End
2014-06-30
Support Year
2
Fiscal Year
2013
Total Cost
$200,431
Indirect Cost
$70,702
Name
University of Rochester
Department
Radiation-Diagnostic/Oncology
Type
Schools of Dentistry
DUNS #
041294109
City
Rochester
State
NY
Country
United States
Zip Code
14627
Nagarajan, Mahesh B; Coan, Paola; Huber, Markus B et al. (2014) Computer-aided diagnosis for phase-contrast X-ray computed tomography: quantitative characterization of human patellar cartilage with high-dimensional geometric features. J Digit Imaging 27:98-107
Yang, Chien-Chun; Nagarajan, Mahesh B; Huber, Markus B et al. (2014) Improving bone strength prediction in human proximal femur specimens through geometrical characterization of trabecular bone microarchitecture and support vector regression. J Electron Imaging 23:013013
Nagarajan, Mahesh B; Huber, Markus B; Schlossbauer, Thomas et al. (2014) Classification of small lesions on dynamic breast MRI: Integrating dimension reduction and out-of-sample extension into CADx methodology. Artif Intell Med 60:65-77
Nagarajan, Mahesh B; Huber, Markus B; Schlossbauer, Thomas et al. (2013) Classification of small lesions in dynamic breast MRI: Eliminating the need for precise lesion segmentation through spatio-temporal analysis of contrast enhancement over time. Mach Vis Appl 24:
Gongvatana, Assawin; Harezlak, Jaroslaw; Buchthal, Steven et al. (2013) Progressive cerebral injury in the setting of chronic HIV infection and antiretroviral therapy. J Neurovirol 19:209-18
Nagarajan, Mahesh B; Coan, Paola; Huber, Markus B et al. (2013) Computer-aided diagnosis in phase contrast imaging X-ray computed tomography for quantitative characterization of ex vivo human patellar cartilage. IEEE Trans Biomed Eng 60:2896-903
Nagarajan, Mahesh B; Huber, Markus B; Schlossbauer, Thomas et al. (2013) Classification of Small Lesions in Breast MRI: Evaluating The Role of Dynamically Extracted Texture Features Through Feature Selection. J Med Biol Eng 33: