The Biomedical Image Analysis Group

The Biomedical Image Analysis Group is a research group devoted to biomedical image processing and analysis, image physics and acquisition, new technologies for data acquisition and reconstruction in magnetic resonance imaging (MRI) and the study of dynamic complex systems with an special focus on the brain.

We also work on new technological developments around the brain that comprise medical imaging, electronics, computer vision and 3D printing.

Our young and enthusiastic group always welcome new ideas and projects in the field of biomedical applications with a highly technological component.

Some research projects

Treating disease by retuning brain network dynamics - Texture analysis on multimodal brain Magnetic Resonance imaging for the early detection of network disturbances and disease biomarkers

Spanish Ministerio de Economía y Competitividad (MINECO), grant ref. BFU2015-64380-C2-2-R

The brain is composed of massively connected elements arranged in modules that form hierarchical networks. The different modules do not operate in isolation; on the contrary, interactions at multiple levels occur producing the characteristic fluctuations of brain activity. Theory shows that connecting stable networks yields structures in which small perturbations in one network are amplified in a cascade of interactions between networks, resulting in frequent catastrophic failures. Recent results from awake human and rodent resting-state fMRI reveal a well-defined connectivity design, characterized by the presence of strategically connected regions (core nodes) which critically contribute to resilience in interacting brain networks. These findings predict that modifying activity in a set of core nodes should drastically alter global patterns of brain activity; which, in turn, raises the possibility that certain brain pathologies could just be the consequence of cascading effects amplifying alterations in normal connectivity that could be moderate in origin. We propose that mapping core nodes and concomitant structural/functional alterations in brain pathology with state-of-the-art network and multimodal texture analysis, respectively, should provide mechanistic understanding of disease and novel therapeutic strategies. The strong multidisciplinary expertise required for this work will be realized in a truly coordinated research plan. First, we will localize core nodes in dynamic resting-state fMRI networks in awake rats and investigate their centrality for large-scale functional connectivity with multiple electrophysiological recordings. Then, we will determine causality between core nodes and brain states by perturbing predefined nodes with an arsenal of available optogenetic and pharmacogenetic tools and deep brain stimulation (DBS). In parallel experiments, we will exploit the combination of MRI modalities hybridizing neuroimaging datasets through multivariate pattern analysis to boost its sensitivity and accuracy to unveil brain dysfunction. Textural feature of brain images will be the input to model-free classifiers to differentiate between brain network states. We will further explore the possibility of combining functional and structural textural information to build efficient disease biomarkers, pushing this technique beyond the state-of-the-art to enhance its diagnostic capabilities. Finally, we will adopt a high-risk high-gain approach to develop network therapy to treat maladaptive brain dynamics. The hypothesis here is that interventions directed to core nodes could serve as access points to control network dynamics and, therefore, represent potential therapeutic targets to retune maladaptive network dynamics. We will test this hypothesis using two well-established models of alcohol use disorder and postulate that the differences between brain networks mediating normal and pathological drive for alcohol are largely mediated by a set of distinct nodes that push the network to a new state (allostasis). Modulating neuronal activity with new-developed DBS protocols directed to brain regions identified in the combined network and texture analysis will need to demonstrate, for a proof of concept, retuning of brain dynamics and improved behavioral output.