Lucia Vaina - US grants

The principal objective of this study is to understand the functional pathways underlying the visual processing of motion in the human brain. The subjects for the study will be a group of normal adults and stroke patients with focal lesions in the occipital, temporal or parietal lobes, whose deficits will be also characterized on neurological, neuropsychological, neuro-ophthalmological and neuroanatomical dimensions. Recent physiological and computational advances in the understanding of primate motion analysis provide the framework for the experimental studies proposed in this project. Two classes of problems in visual motion processing will be addressed: (1) detection and measurement; and (2) the use of motion measurements for visual tasks such as the separation of figure from the background or for the derivation of three dimensional structure, for example. The second objective of the study is to learn whether deficits of motion and form processing tend to be associated with functionally and structurally distinct systems in the brain. The third objective is to identify the specific motion deficits that may be correlated with visual-spatial impairments and with poor performance on neuropsychological tasks designed for assessing right hemisphere functions. This might provide the basis for the remediation programs of visuo-spatial deficits. We are specifically interested in the correlations of visual motion deficits and stereopsis, contrast sensitivity, acuity and form.

The principal objective of this study is to understand the functional pathways underlying the visual processing of motion in the human brain. The subjects for the study will be a group of normal adults and stroke patients with focal lesions in the occipital, temporal or parietal lobes, whose deficits will be also characterized on neurological, neuropsychological, neuro-ophthalmological and neuroanatomical dimensions. Recent physiological and computational advances in the understanding of primate motion analysis provide the framework for the experimental studies proposed in this project. Two classes of problems in visual motion processing will be addressed: (1) detection and measurement; and (2) the use of motion measurements for visual tasks such as the separation of figure from the background or for the derivation of three dimensional structure, for example. The second objective of the study is to learn whether deficits of motion and form processing tend to be associated with functionally and structurally distinct systems in the brain. The third objective is to identify the specific motion deficits that may be correlated with visual-spatial impairments and with poor performance on neuropsychological tasks designed for assessing right hemisphere functions. This might provide the basis for the remediation programs of visuo-spatial deficits. We are specifically interested in the correlations of visual motion deficits and stereopsis, contrast sensitivity, acuity and form.

9528079 Vaina This project is a Small Grant for Exploratory Research. The feasibility of new approaches to vision applied to recognition problems of use to the visually impaired is investigated. These approaches combine coarse, easily obtainable, visual data with novel adaptive learning algorithms to recognize features such as moving objects and terrain irregularities. The results of this research project will be feasibility study as to the robustness and practicality of this approach in developing a "magic hat" - an aid to the visually impaired consisting of a sensor and computation ring that would be worn by users as a hat and which adaptively warn the user of hazards. ***

The goals of the POWRE grant are two-fold: first, to assist the investigator in acquiring expertise in the newly emerging technology of functional magnetic resonance imaging (fMRI), and second, to combine that expertise with her background and strength in visual psychophysics and computation modeling to study mechanisms underlying visual motion perception and their modulation by learning. The major training components of the grant include the investigator's participation in tutorials, seminars, NMR training sessions and course work, and collaborative research and developing skills in fMRI data analysis, developing fMRI specific activation paradigms, and methods of representation of fMRI activations. The broad scope of the research component of this proposal, to be carried out at the Nuclear Magnetic Resonance Center at Massachusetts General Hospital (MGH-NMR), is on combining fMRI and psychophysics to study functional plasticity in the normal adult human visual system, specifically the mechanisms and neural circuitry underlying learning motion discrimination in humans. Cortical fMRI signal changes (in both the cerebral and cerbellar cortices) can be reliably measured, and will be used to ask specific questions about practice and stimulus attributes-related changes in human brain functional anatomy. The approach of the proposed experiments will also serve as validation for the learning technique as a powerful psychophysical tool to investigate specificity and dynamics of the neural circuitry underlying visual perception and its modulation by attentional mechanisms. The POWRE funding mechanism will provide the investigator, already a senior scientist and educator specialized in human vision, the opportunity to bring that knowledge to bear in the design and interpretation of neurobiologically relevant imaging experiments. It will also give her the opportunity to become proficient in fMRI techniques and to gain additional knowledge that she will subsequently apply in her research and teaching. The MGH-NMR is a superb training site for accomplishment of these goals. The results of this training will significantly increase the candidate's visibility in the neuroscience and human brain mapping community. Also, it will allow her to play a key role in the efforts of the Biomedical Engineering Department at her home institution to develop a strong neurosciences track in the graduate program in biomedical engineering.

PI: VAINA
Abstract

The objective of this proposal is to investigate computationally a biologically plausible model for learning invariance. Invariant recognition is a fundamental capacity of perceptual systems, which makes it possible to recognize visual objects under different viewing conditions, such as changes in the relative position of the object in the visual field, changes in distance, viewing direction, illumination, and also under shape deformations. The purpose of this study is to extend and generalize the shift invariance algorithm and incorporate a broader class of invariances, such as size and rotation in depth and combine it with a learning algorithm to achieve solutions to novel types of computational problems in biological vision.

A major focus of this research will by on computational studies of biological visual problems for which there is ample neurophysiological and psychophysical data. Two important and novel characteristics of the models to be developed are: 1) invariance is achieved gradually in a series of processing stages, and 2) a simple unsupervised learning mechanism is sufficient for connecting units in a neural network in a way that results in invariant recognition. This approach is broader than previous approaches and therefore riskier. However, if it is shown to be correct, then it will offer a uniform account of multiple aspects of invariant perception and allow the development of powerful learnable mechanisms for computer vision systems.

DESCRIPTION (Adapted from applicant's abstract): A fertile convergence of anatomical, electrophysiological, psychophysical, functional imaging and computational approaches has made visual motion one of the most active areas of research with a number of important well-defined problems that are yielding to rigorous investigation. We expect that the outcome of the research program proposed here will contribute in important ways to our understanding of the functional architecture of the human visual motion system by a) elucidating the nature and organization of the mechanisms involved in the visual guidance of navigation, and b) investigating whether a particular anatomical region is computationally necessary for specific visual functions. The method by which we will address this is through studying the performance of neurological patients with discrete lesions on talks specifically designed to address the cortical mechanisms believed to underlie visual navigation. How does locomotor guidance depend on optic flow, 3D scene structure, and the recognition of landmarks? Are motion and stereo integrated at different stages in this process? We have previously developed a psychophysical paradigm of computer generated motion stimuli designed to critically analyze motion. Within this paradigm we shall now develop a new set of tasks designed to examine the effect of discrete brain lesions in neurological patients on (1) the perception of speed and direction of complex motion patterns, discrimination of the center of motion, heading perception; (2) the interaction between stereopsis and global motion and how different cues for 3D perception are processed; (3) perceptual-motor actions. Progress in understanding the neural substrate and the mechanisms underlying specific deficits in motion for navigation will provide important clues for how perception is linked to action. The correlation of neuroimaging, neuro-ophthalmological, neurological and neuropsychological examinations with patients' performance on the psychophysical tasks of visual navigation described in this proposal will be useful for devising diagnostic and mediation strategies.

Stroke continues to be the leading cause of long-term disability among adults and its prevalence will continue to rise as the population ages. Developing analytical strategies for improving quality of life and independence following stroke are of tall importance. For this endeavor to be successful, a critical and necessary step is to understand the neuro-scientific basis of the underlying mechanisms. and to integrate this knowledge with the translational science of rehabilitation. This is what we propose to do. Over the next 2 years, we propose a quantitative and multifaceted research program that integrates neurology, neuroscience, psychophysics and brain imaging to study the visual mechanisms directly relevant to visually guided behavior and the effects of brain lesions (from stroke) on patients'ability to carry out everyday activities. We employ a hypothesis-driven approach to provide a solid scientific basis for integrating basic neuroscience with the translational science of recovery and rehabilitation. We have three Specific Aims: Aim 1. To characterize the mechanisms for recovering an object's 3D motion during selfmotion through the environment. We test the hypothesis that recovery of object trajectory during selfmotion requires the visual system to account for the induced motion of stationary objects in the scene. Aim 2. To examine the functional organization of visual motion processing for collision detection in the human brain and test the sufficiency of alternate visual cues and behavioral strategies when primary mechanisms are impaired. We test the hypothesis that collision detection is mediated by a distributed network of relative motion mechanisms that support obstacle avoidance and investigate the use of alternative strategies for recovering obstacle avoidance following stroke. Aim 3. To determine the relationship between performance on early visual motion tasks and activities of visually guided navigation. The experimental results obtained from patients on the screening tests batteries will be analyzed across the patient population using quantitative statistical analysis (k-means clustering) to identify clusters of early visual motion and attention tests that diagnostic of selective deficits in stroke patients. The work we propose to carry out over the next two years will elucidate the neuronal substrate of essential visual mechanisms involved in visually guided navigation and explore alternate cues that patients impaired on these tasks may use for coping with specific aspects of their environment. Furthermore, we expect that the outcome of the research proposed here will have a significant clinical impact on future design of targeted neurorehabilitative therapies for functional recovery from deficits of visually guided navigation and mobility.

1545668(Vaina)

The PI will develop a novel real-time, magnetoencephalography based neurofeedback method designed to rehabilitate deficits on executive functions such as decision making, switching attention, search, and error correction. Benefits of this work encompass scientific understanding, methodological trailblazing, with significant impact on both dimensions and an opening of a new, more optimistic, era in the neurorehabilitation of executive functions which will improve the quality of life of impaired individuals. It is important to emphasize that although this application focuses on training executive functions in the aging population, principles derived via the proposed training, will lead to neurorehabilitation of executive functions deficits in other populations such as stroke, traumatic brain injuries, and in individuals afflicted by psychiatric or developmental disorders. Undergraduate and graduate students in science and engineering will benefit from hands-on involvement in the project, providing them with an understanding of how engineering methods can contribute to human neuroscience of cognitive rehabilitation.

This project has two related goals. First it aims to develop a novel and transformative neuro-therapeutic intervention, State-modulating neurofeedback (smNFB), which will provide rapid modulation of cortical activity targeting training dynamic brain functions typical of Executive Functions (EF). The expected outcome of the training is increased speed to switch spatial attention in the trained task. Novel algorithms and simulations will be implemented to assure that the smNFB method effectively provides real time neurofeedback (rt-NFB) useful as a neurotherapeutic protocol. This method is innovative and, if successful, will significantly contribute to the success of neurorehabilitation targeting deficits in EF. The second goal is to monitor changes in cortical activation and dynamic connectivity as a function of smNFB training. It is expected that smNFB training will affect the strength of activation and the timing of connectivity in the frontoparietal network involved in executive control of attention. Functional connectivity between candidate cortical areas will be assessed through an extension of the dynamic Granger Causality methods developed in the PI laboratory, which measures how the magnitude of directed functional connectivity between cortical areas evolves temporally during training. Thus the second innovation of this project is that it provides a multifaceted quantitative neuroscientific validation of the proposed smNFB method and how it affects the spatiotemporal patterns of cortical processing. A third innovation of this project is a unique way for assessing efficacy of the neurofeedback cue within subject, which is highly relevant for efficacy testing of all neurofeedback paradigms in patient populations whose performance and response to rehabilitation are often very variable and inconsistent. The project will provide open source software tools for the scientific community, following gold-standard guidelines for sharing software (March 2014 Editorial of Nature Methods). These tools will be integrated in the Brain Storm toolbox which is a free and widely used MEG/EEG data analysis software package written in Matlab as will be the algorithms developed under the proposed research.