This thesis addresses the computational modeling of brain mechanics and fluid flow in image-reconstructed patient-specific geometries, using a poroelastic-Stokes framework to represent the interaction between deformable parenchyma and free-fluid regions such as the ventricles and cerebrospinal spaces. Motivated by clinical problems where deformation and fluid transport are tightly coupled, like traumatic brain injury, hydrocephalus, and cerebral edema, this work aims to advance the state of the art of meshing and simulation tools. The central challenge addressed is the translation of clinical images into simulation-ready domains while preserving anatomical fidelity. Converting MRI/CT volumes into volumetric meshes requires careful segmentation, surface repair, and mesh quality control; simultaneously, the multiphysics coupling and near-incompressible tissue behaviour require high geometrical accuracy and robust and efficient computational methods. To meet these challenges I have developed an end-to-end pipeline that integrates image preprocessing and reconstruction strategies, volumetric mesh generation, a poroelastic-Stokes mathematical description, and finite element discretisations designed for the coupled problem by combining several software libraries. The results highlight both the promising upcoming and current limitations of multiphysics brain modelling and point to future work in validation against experimental and clinical measurements.
Questa tesi realizza mesh paziente-specifiche per la modellazione di meccanica e fluidodinamica cerebrale, sviluppando una pipeline end-to-end che integra strategie di preprocessing e ricostruzione delle immagini adatte alla generazione di mesh volumetriche. Il dominio tridimensionale ottenuto viene utilizzato come base per un modello matematico accoppiato di interazione fluido-struttura porosa per descrivere la connessione tra il parenchima cerebrale deformabile e le regioni a fluido libero - per esempio i ventricoli cerebrali e gli spazi occupati dal liquido cerebrospinale - e una discretizzazione agli elementi finiti pensata per il problema accoppiato e implementata tramite diverse librerie in Python. Motivata da problemi clinici in cui deformazione del tessuto e trasporto di fluido cerebrospinale sono fortemente interconnessi (trauma cranico, idrocefalo, edema cerebrale), la ricerca si propone di avanzare gli strumenti di generazione di mesh e di simulazione. Il problema principale affrontato riguarda la conversione di immagini cliniche in domini computazionali: trasformare i volumi ottenuti mediante risonanza magnetica e tomografia computerizzata in mesh volumetriche, mantenendo la massima fedeltà anatomica possibile, richiede segmentazioni accurate e controlli stringenti sulla qualità della mesh. I risultati evidenziano sia le promettenti potenzialità, sia i limiti attuali della modellazione multifisica cerebrale e indicano la necessità di lavori futuri orientati alla validazione mediante misurazioni sperimentali in-vivo e dati clinici.
Computational fluid-poromechanics modelling of the brain in three-dimensional patient-specific geometry
Vanini, Tommaso
2024/2025
Abstract
This thesis addresses the computational modeling of brain mechanics and fluid flow in image-reconstructed patient-specific geometries, using a poroelastic-Stokes framework to represent the interaction between deformable parenchyma and free-fluid regions such as the ventricles and cerebrospinal spaces. Motivated by clinical problems where deformation and fluid transport are tightly coupled, like traumatic brain injury, hydrocephalus, and cerebral edema, this work aims to advance the state of the art of meshing and simulation tools. The central challenge addressed is the translation of clinical images into simulation-ready domains while preserving anatomical fidelity. Converting MRI/CT volumes into volumetric meshes requires careful segmentation, surface repair, and mesh quality control; simultaneously, the multiphysics coupling and near-incompressible tissue behaviour require high geometrical accuracy and robust and efficient computational methods. To meet these challenges I have developed an end-to-end pipeline that integrates image preprocessing and reconstruction strategies, volumetric mesh generation, a poroelastic-Stokes mathematical description, and finite element discretisations designed for the coupled problem by combining several software libraries. The results highlight both the promising upcoming and current limitations of multiphysics brain modelling and point to future work in validation against experimental and clinical measurements.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_12_Vanini_Tesi.pdf
accessibile in internet per tutti
Descrizione: Tesi
Dimensione
34.92 MB
Formato
Adobe PDF
|
34.92 MB | Adobe PDF | Visualizza/Apri |
|
2025_12_Vanini_Executive Summary.pdf
accessibile in internet per tutti
Descrizione: Executive summary
Dimensione
5.44 MB
Formato
Adobe PDF
|
5.44 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/247146