About BRIDGE-AI

Neurological disorders such as stroke, dementia, epilepsy and multiple sclerosis are collectively the leading cause of disability, and the second highest cause of death in the world. The enormous economic and societal impact of these disorders is projected to increase even further as more people reach ages at which neurological diseases are more prevalent — particularly concerning to an aging European society.

In the clinical treatment of chronic brain disorders, therapy selection is driven by the level of disease activity and progression that can be observed. Medical imaging plays a prominent role in the monitoring of patients and in evaluating the effectiveness of their treatment: By comparing images of a patient with corresponding images taken at earlier time points (aka longitudinal imaging), subtle changes in the structure and the function of the brain can often be detected long before clinical symptoms deteriorate.

Visual assessment of longitudinal images by human experts is fundamentally limited by its inefficiency, its inability to quantify diffuse changes, and its intra- and inter-rater variability. Therefore, to better predict disability progression and treatment response in brain disease patients, there is an urgent need for new computational tools that can reliably help clinicians characterize temporal changes in their patients.

Against this backdrop, the BRIDGE-AI project aims to attain the following research objectives through the brilliant minds of 15 doctoral candidates (DCs):

  • To develop new methods for longitudinal monitoring of brain disorders that cover the entire lifespan
  • To ensure the trustworthy behavior of the proposed methods so that they work robustly across imaging protocols and scanners used at different institutions
  • To focus explicitly on methods that support rather than supplant human experts in their work
  • To leverage the network to enable large-scale validation and dissemination of the developed tools.

The research program in the BRIDGE-AI project is glued together by a common focus on these four research objectives and on open-source software development, and is organized in the following three work packages:

  • WP1: Longitudinal segmentation in diseased populations (voxel-level analysis)
  • WP2: Subject-level tracking and future outcome prediction (patient-level analysis)
  • WP3: Radiological phenotyping of temporal trajectories (group-level analysis).