Master of Science
During everyday behaviours, the brain shows complex spatial patterns of activity. These activity maps are very replicable within an individual, but vary significantly across individuals, even though they are evoked by the same behaviour. It is unknown how differences in these spatial patterns relate to differences in behavior or function. More fundamentally, the structural, developmental, and genetic factors that determine the spatial organisation of these brain maps in each individual are unclear. Here we propose a new quantitative approach for uncovering the basic principles by which functional brain maps are organized. We propose to take an generative-discriminative approach to human brain mapping, with the fundamental idea that, if we understand the underlying principles that organises brain activity maps, we should be able to generate sets of artificial maps that are indistinguishable from real ones. Different generative models are tested by a series of adversarial, classifier models, ranging from linear classifiers based on specific marginal statistics, to a full convolutional neural network. We apply our new framework to a collection of measured finger activity maps measures with fMRI in the human sensori-motor cortex (N=50). To account for characteristics of the brain maps that depend on the specific measurement process (spatial resolution of fMRI, signal-to-noise, etc) we supplemented the generative process with a measurement model. Initial results clearly demonstrate that the matching of simple marginal statistics (covariance and smoothness of activity pattern) is sufficient to fool the human eye, but not a more systematic machine learning approach. The proposed evaluation framework therefore opens up a pathway for discovering specific characteristics of brain activity maps that are important to explain function or individual differences.
Summary for Lay Audience
Our brains are structured differently. Each individual has a unique brain structure that is shaped by multiple factors, including genetic and developmental factors. Additionally, different parts of our brain work differently and are structured in a way that helps them specialize in their specific function/s. The functioning of the brain, and the resultant activity in different regions of the brain can be recorded and measured using a variety of imaging techniques, including fMRI. These patterns of activity (activity maps) vary widely across individuals and are important in understanding the structure and the functioning of the brain.
My thesis introduces a novel approach towards brain mapping and is aimed at determining the principles that characterize the organisation of the brain activity maps across individuals. Discovering these underlying principles is important, as they have the potential to relate to individual differences in function, or to characterize patient groups with dysfunction in specific domains. This research specifically focuses on understanding the organization of activity patterns in the sensorimotor cortex of the brain.
For understanding these principles, we developed generative models to generate artificial maps based on a dataset of real maps. The generative models were designed to test various hypotheses about the structure of the brain activity patterns. For evaluating the generative models, we developed a comprehensive evaluation framework that is adaptable to any generative model under test. The evaluation framework provides insights into the points of strengths and weaknesses in a generative model. The specific patterns of strengths and weaknesses then guide the improvement of the corresponding generative model/s. The generative model and the evaluation framework hence act as adversaries where the generative model tries to fool the evaluation framework by generating fake activity maps.
This research hence contributes to the broader scientific in multiple ways – as the proof of insufficiency of known organizational principles as the sole organizational factors in the activity maps, as the proponent of a new generative-discriminative approach towards brain mapping; as a descriptor of a modelling and measurement simulation process for generating artificial brain maps; and as a source of curated data for further research and development.
Wadhwa, Deepanshu, "A Generative-Discriminative Approach to Human Brain Mapping" (2021). Electronic Thesis and Dissertation Repository. 8092.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.