PHD2020-14
Shape recognition and 3D visualization of neural circuitries from brain tissue clearing
Host laboratory and collaborators
Djamal Merad / LIS / djamal.MERAD@univ-amu.fr
Valéry Matarazzo / INMED / valery.matarazzo@inserm.fr
Christophe Pellegrino / INMED / christophe.pellegrino@inserm.fr
Abstract
Neural networks connectivity is crucial to elucidate how the brain process information during behavioral tasks. The project we propose is to dissect out the anatomical link that exists between the prefrontal cortex and the hippocampus in the frame of social behavior in physiologic and pathologic conditions.
Although a new momentum in neural circuitry visualization has appeared with recent development of new tissue clearing methods, a combined effort between neurobiologists and computer scientists is now required for processing, modeling and analyzing.
The retained candidate with background in informatics and image processing will model the prefontal cortex to hippocampus interaction map using bank of images obtained from brain tissue clearing and light sheet microscopy. One of the originalities of our work is to allow comprehension from the analysis and the recognition of the acquired information until the problems of visualization of the modeled objects.
Keywords
3D Modeling, shape recognition, tissue clearing, image analysis, Neuroscience
Objectives
Our goal is to map the brain connection between the hippocampus and the prefrontal cortex in physiologic and pathologic conditions. Our project will advance the state of the art of 3D brain visualization systems by providing a truly multimodal 3D visualization system enabling true scientific visualization of complex brain datasets. The software developed will also facilitate image segmentation and classification and 3D point clouds information. The candidate will perform brain reconstruction using tissue-clearing techniques.
Proposed approach (experimental / theoretical / computational)
In order to reconstruct brain circuitry, we will segment and classify automatically the 3D point clouds obtained by neuroanatomic images. Many methods exist for the recognition of objects in scenes. They incorporate the implicit knowledge held by their designers about the type of searched objects. However, this kind of approaches had been criticized due to its lack of generality and the problem of representation and acquisition of prior knowledge. However, recent advances such as ontology, permit to address some of these criticisms by formalizing the knowledge of a particular field in a coherent and consensual manner. We propose a novel approach to provide methods and tools that use the knowledge of the field and exploit the complementarities of data (2D images, 3D points...) to automatically extract and recognize elements in the scene. We will set up an ontology integrating complex data describing a 3D model of neural circuitry. Alignment of this model in the scene will allow the identification and localization of all occurrences of this model.
Interdisciplinarity
This project groups together researchers in computer science specialized in image processing and shape recognition (LIS research laboratory) and neurobiologists expert in brain connectivity, circuitry remodeling and brain imaging from clearing tissue (INMED). Both laboratory are already equipped with tagging neuronal circuitry and tissue clearing methods, light sheet microscope and computers for big data analyses.
With the expertise of the LIS, we will be able to develop software enabling to model neuronal circuitry during development and how these circuitries are remodeled in pathologic conditions.
This project will also train the PhD candidate to develop an interdisciplinary expertise in computer science applied to neuroanatomical imaging.
Expected profile
We expect a candidate with background in informatics and image processing for the recognition of objects. The candidate will be involved from tissue clarification and image acquisition to development of imaging software.