/AI-based Image Analysis and Registration for Functional Ultrasound based Brain Activation Detection and Mapping

AI-based Image Analysis and Registration for Functional Ultrasound based Brain Activation Detection and Mapping

PhD - Gent | More than two weeks ago

Decode how the brain works using cutting-edge AI-driven Functional Ultrasound technology.

Understanding how networks of neurons in the brain convey and process information requires recording neural activity with high spatial and temporal resolution. Conventional imaging methods such as fMRI or PET are cumbersome, expensive, and cannot be applied to freely moving subjects. Functional UltraSound Imaging (fUSI) is a novel ultrasound-based technology that allows us to monitor brain-wide neuronal activity through changes in local hemodynamics (activated brain regions require higher blood flow than non-activated regions). Coupled with video recordings of small animal experiments, the setup allows the researchers to map the brain activity to distinct movements/behavior of laboratory animals in resting conditions (brain connectivity), evoked activity, and during tasks (event-related). The produced data are complex volumetric datasets (3D-in-time up to 100 μm3 voxel size) whose analysis is particularly challenging due to the high dimensionality. Developing and testing algorithms and tools for large-scale data mining becomes necessary. Specifically, analysis of complex 3D-in-time data requires: (1) accurate automatic segmentation and centerline extraction of smallest brain blood vessels in fUSI (often obstructed with noise and unequal contrast), (2) AI-based 3D-in-time image registration between time instances in order to track the changes in particular vessels over time, and (3) near-to-real-time processing and visualization to allow for live laboratory experiments.

Neuro-Electronics Research Flanders (NERF) group of IMEC-VIB-KU Leuven has vast experience developing the fUSI technology and conducting animal studies to detect and link brain activity to animal behavior. NERF team recently developed volumetric (3D in time) fUSI (vfUSI) for imaging large-scale functional networks in real-time and at high spatial resolution (up to 100 μm3 voxel size). Such an abundance of data requires accurate and automatic image processing techniques. The Image Processing and Interpretation (IPI) group of IMEC has been working on several biomedical and medical image processing and analysis applications and has a long-standing experience in the field. The in-house developed domain-specific programming language “Quasar” and IDE “Redshift” are used to facilitate rapid prototyping of complex image and real-time video processing algorithms (http://gepura.io/quasar/).

We are looking for a motivated Ph.D. candidate interested in pushing the fUSI technology to the forefront of functional brain imaging. You will:

1.         Develop novel brain vessel segmentation and centerline extraction methods to allow for efficient and accurate brain activation detection.

2.         Develop novel AI-based registration methods for tracking the activation of vessels and brain regions over time.

3.         Design methods for efficient parallel data processing to allow for near-to-real-time assessment of brain region activation (“Quasar” and “Redshift”, http://gepura.io/quasar/).
 

AI-based Image Analysis and Registration for Functional Ultrasound based Brain Activation Detection and Mapping


Required background: Computer Science (Image Processing and AI)

Type of work: 70% modeling, 20% experimental, 10% literature

Supervisor: Bart Goossens

Co-supervisor: Alan Urban

Daily advisor: Danilo Babin

The reference code for this position is 2023-076. Mention this reference code on your application form.

Who we are
Accept marketing-cookies to view this content.
Cookie settings
imec's cleanroom
Accept marketing-cookies to view this content.
Cookie settings

Send this job to your email