WP2 Novel imaging disease biomarkers & methods for model personalization

Rationale:
This WP will focus on aspects of model personalization and biomarker extraction that require synergistic developments in MR image acquisition and computational imaging. The latter will be tackled by applying techniques from image computing (including model-based methods), inverse problems and scientific visualization to enable estimation and exploration of model parameters from medical images. USFD, PRH, INSERM, VTT and ICL will apply in-depth knowledge on image computing and computational physiology in these developments and act as a bridge between the parameter estimation associated with the modeling activities in WP4-6 and advanced MR image acquisitions. Into this context, ETHZ, USFD, PRH and INSERM will provide their expertise in advanced MR pulse sequence design and MR data processing for quantification of venous and arterial blood flow, CSF flow, diffusion anisotropy, venous vascular alterations, plaque/micro-bleed density (MR susceptibility) and for Magnetic Resonance Elastography (MRE). Highly accelerated MR data acquisitions in conjunction with advanced parameter encoding strategies will be implemented to permit data acquisition and quantification in clinically feasible exam times. First-stage processing tools will be implemented to derive quantitative flow vector fields, fibre tract geometries and diffusion compartments, magnetic susceptibility related parameters, as well as viscoelastic and poroelastic tissue properties. These parameters will serve as input to advanced modeling as developed at ASD, UOXF and EMC (WP4-6)
Objectives:

Firstly, the development of advanced Magnetic Resonance (MR) imaging methods and protocols to map time-resolved venous, arterial and CSF flow, brain fibre architecture, visco- and poroelasticity of brain tissue and structural alteration of venous structures. Secondly, the development of computational imaging tools for mechanistic model personalization and extraction of novel biomarkers for dementia quantification. These data and methods will be added to conventional MR readouts as acquired in WP1 and existing tools in WP3, and will feed IT software that enables efficient and user-friendly construction and personalization of computational physiology models. This work package will require synergistic developments both in terms of image acquisition and computational imaging. The package broadly comprises the following image-related tasks:

  • development of advanced MR image acquisition and reconstruction techniques
  • implementation of first-stage data processing to derive MR encoded parameters
  • extraction of anatomical computational subdomains from anatomical MR images
  • extraction of tissue class distributions and their material properties
  • development of model- and image-based biomarkers for dementia classification
  • translation of the methods and their integration into clinical prototypes
Activities:
  • Intracranial blood/CSF velocity vector imaging & brain fibre architecture imaging
  • In-vivo brain imaging of viscoelastic and poroelastic properties
  • In-vivo brain MR susceptibility quantification of iron containing plaques and venous structures
  • Shape models of cortical and sub-cortical brain structures from MR images
  • Multi-sequence/multi-modal alignment of brain scans
  • Statistical diffusion modelling and tractography of brain fibre bundles
  • Multi-poroelasticity model parameter identification from MRE data
  • Quantification of vascular dementia from MR images
  • Normative models of brain tissue image intensity in MR and PET/SPECT based on biophysical models and models of imaging physics