Patient-Speci�c Velocity Boundary

Conditions from Phase Contrast

Magnetic Resonance Imaging

Andrea Torti

Università degli Studi di Pavia

February 26, 2014

Supervisor: Simone Morganti, PhD

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 1 / 22

Outline

Computational Fluid Analysis in the Biomedical Field

Goal of the thesis

From PC MRI data to patient-speci�c velocity pro�les

Conclusions and Future Works

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 2 / 22

Outline

Computational Fluid Analysis in the Biomedical Field

Goal of the thesis

From PC MRI data to patient-speci�c velocity pro�les

Conclusions and Future Works

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 3 / 22

Outline

Computational Fluid Analysis in the Biomedical Field

Goal of the thesis

From PC MRI data to patient-speci�c velocity pro�les

Conclusions and Future Works

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 4 / 22

Outline

Computational Fluid Analysis in the Biomedical Field

Goal of the thesis

From PC MRI data to patient-speci�c velocity pro�les

Conclusions and Future Works

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 5 / 22

Computational Fluid Analysis

Computational Fluid Analysis (FSI/CFD) is:

non-invasive

potentially very accurate

predictive

Therefore very useful in the Biomedical �eld

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 6 / 22

Computational Fluid Analysis (FSI/CFD)

A widely investigated issue for practical purposes:

Gerbeau and Vidrascu, 2003 �> Algorithms for FSI

Papaharilaou et al., 2006 �> FSI for Abdominal Aortic Walls Stress

Bluestein et al., 2008 �> FSI for Abdominal Aortic Aneurysm

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 7 / 22

FSI/CFD Recipe

...What is needed?

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 8 / 22

Geometrical Domain - Patient-Speci�c Approach

Nealand and Kerckho�s, 2009 �> Progress in Patient-Speci�c Approaches

Auricchio et al., 2014 �> CFD for TEVAR evaluation

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 9 / 22

Aim of the thesis

GOAL

De�nition of a time and space-dependent Aortic In�ow using Patient-Speci�c Data from PhaseContrast Magnetic Resonance Imaging

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 10 / 22

Available Data - PC MRI

PC-MRI: physical principles

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 11 / 22

PC MRI

Two kinds of data

Unlike standard MRI, PC-MRI also employs information from phase maps

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 12 / 22

Clinical Data

30 phase maps (ascending aorta slice) extracted via PC MRI at I.R.C.C.S. San Donato, Milan

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 13 / 22

Image Cropping and Segmentation

1) Rectangular, Automatic Cropping with ImageJ

2) Elliptical, Semi-Automatic Segmentation with Matlab

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 14 / 22

From Images to Matrices

Once in Matlab, each image is related to a matrix, whose cells contain values in Houns�eld units(HU, from 0 to 255, measuring tissue density).

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 15 / 22

From Houns�el Units to Velocities

In PC-MRI, mid-gray represents steady tissues (HU0 = 127)

HUmax = 255 �> MRI Venc (in this case, 200 cm/s)

Being R = Venc/HUmax , we can use the following relation:

v(i,j) = (HU(i,j)-HU0)R

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 16 / 22

Mean Velocities Calculations

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 17 / 22

Comparing Datasets

Our Data vs Machine-provided Data

In blue, our plot representing mean velocity vs time, compared to the data provided by I.R.C.C.S.San Donato (in red)

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 18 / 22

Results

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 19 / 22

Conclusions

In the present work:

literature review on computational �uid analysis

collection of patient-speci�c PC MRI data

elaboration of the provided data to obtain time- and space-dependant velocity pro�les

Obtained patient-speci�c aortic in�ow from PC MRI is in good agreement with machine-provideddata

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 20 / 22

Future Work: Data interpolation

Data are represented on a �ner grid (but still discrete!)

We would need an interpolant function �> (LIFE V)

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 21 / 22

Patient-Speci�c Velocity Boundary

Conditions from Phase Contrast

Magnetic Resonance Imaging

Andrea Torti

Università degli Studi di Pavia

February 26, 2014

Supervisor: Simone Morganti, PhD

Andrea Torti (unipv) Boundary Conditions from PC-MRI February 26, 2014 22 / 22