CT Brain Perfusion Studies

8/14/2019 CT Brain Perfusion Studies

http://slidepdf.com/reader/full/ct-brain-perfusion-studies 1/28

CT Brain Perfusion Studies

David Rodríguez González

Pedro Santiago del RíoInstituto de Física de Cantabria (UC-CSIC)

Enrique Marco de LucasHospital Universitario Marqués de Valdecilla

EGEE Industry Day, Bratislava, Slovakia, 19 September 2007



EGEE ID. Bratislava 19/09/2007 2

Outline

Motivation

Infarct location Perfusion and hemodynamical maps

Int.eu.grid integration

Future Work




Introduction

Collaboration between IFCA (CSIC-UC) and

Hospital Universitario Marqués de Valdecilla atSantander (HUMV) radiologists.Previous collaboration

Development of a tool for assisted diagnosis forbrain stroke using hemodynamic parametersmaps

Vendor independent parameters calculationtesting several methods

Integration in int.eu.grid






WHAT’S STROKE?




STROKE

One million strokes occurring per year in the EuropeanUnion.

NINDS trial demonstrated usefulness of thrombolytictreatment of stroke in selected patients rescuing theischemic penumbra.

Thrombolytic treatment has an associated risk of cerebral hemorrhage up to 20%.




WHAT’S STROKE?




Core: part of the ischemic region that isirreversibly injured

Penumbra: area of the brain underperfused andin danger of infarcting.

NECROSIS

NECROSIS

PENUMBRA




STROKE

CT perfusion plays a major role by demonstratingsalvageable brain tissue and extension of cerebral

core infarction.

And… WHAT’S CT PERFUSION??







ANALYSIS

Attenuation proportionalto blood (= contrast) in

brain tissue.

Time attenuation curve of

reference:-artery (ACA, MCA)

-vein




Deconvolution (MTT)

Parametric maps

CBV = AUC pixel parench/ AUC pixel artery

CBF= CBV / MTT




Vascular pathology

- Acute ischemic stroke

- Chronic ischemia

- Vasoespasm

Brain tumours

MAIN CLINICAL APPLICATIONS




¿salvageable cerebral tissue??

NECROSIS

NECROSIS

(DWI/CBV)

PENUMBRA (CBF)

NECROSIS

NECROSIS




Infarct location application Using the cerebral blood flow and the mean transit time

generate a brain map showing:

The infarct coreThe ischemic penumbra

Implemented in Java (ij library for the images)

Using input from GE Perfusion applicationWill use also our maps when finished

The program fed with the parameter maps automaticallycalculates an output map

The medic can change the criteria

Also can define a good tissue region to use as reference






CT Perfusion For DCE (Dynamic

Contrast Enhancement)

imaging using CT (aka CTPerfusion) a sequence ofimages (45) at the samelocation are taken in a

given interval.Concertation Time Curves

The objective is to getthe three parameters that

are related by the centralvolume principle

This is done voxel by voxel






Deconvolution Problem

We want to obtain F and R(t)

Ctiss and Cart are the tissue andartery concentrations R(t) is the tissue residue

function and it is used tocalculate the MTT

Numerical deconvolutionprocess is very sensitive tonoise in the measured data

Inherently ill-conditionedproblem

[ ])(*)()(

)()()(0

t Rt C F t C

d t Rt C F t C

art tiss

t

art tiss

=

−= ∫ τ τ

)( / )()(

)()()(

wC wC w R

w RwC wC

art tiss

art tiss

=

=

If using Fourier Transform




Singular Value Decomposition Discretize the convolution

integral equation

System of linear equations

A is a nxn matriz and xand b are vectors

Singular ValueDecomposition (SVD)

b x A =⋅

)(ii

t RF x ⋅=

021

1

≥≥≥≥

=⋅⋅= ∑=

n

n

i

T

iii

T

sss

vsuV S U A

K




Regularization Methods TSVD

Singular Value

Decomposition (SVD)Threshold or truncationindex

The smaller singularvalues are eliminated

Limits the effects of noiseC implementation usingSVDLibC

Java using JAMA

Tikhonov RegularizationPrototype in Matlab

Java using JAMA




Selection of the regularization

parameter Picard plots to estimate

Currently using the L-

Curve method




Interactive European Grid(http://www.interactive-grid.eu)

Projectacronym

int.eu.grid

Contractnumber031857

Instrument

I3

Duration2 years

may ´06-april ´08

“providing transparently

the researcher’s desktop

with the power of a supercomputer,

using distributed resources”

http://www.interactive-grid.eu

Coordinator: CSIC, Jesús Marco, IFCA, Santander, SPAIN [[email protected]]




Project Vision

Distributed Parallel (MPI)Interactive Computing andStorage at the Tera level

Supporting Virtual Organizationsat all levels: setup, collaborative

environment, grid enhancementof applications, execution andmonitoring tools.

User Friendly Access through aGrid Interactive Desktop withpowerful visualization and real

simulation steering in real time




Single sign-on /authorisation

Platform

independent Batch jobs MPI jobs Running interactive

applications using java plugins or VNC

Monitoring gridapplications Flexible Application

framework User profile

management Easy application add

on Local and grid file

management Desktop tools: Job Wizard

Job Monitor

Application Container andPlugin

Migrating Desktop

GridFTP Commander

User Profile Manager

Private Storage

Management VNC/SSH console




Interactivity

CrossBroker

ComputingElement

JDLJob SubmissionServices

Roaming Access Server

gLogin

Simulation+visualisation WorkerNode

master

SimulationResults30GB

Storage Element

Java-based GVid

User

results

…

WorkerNodeslave

WorkerNode slave

WorkerNode slave

input data

MigratingDesktop




Integration in int.eu.grid

The TSVD is already running in int.eu.gridWorking on Tikhonov regularization

The infarct location Java application would beincluded in a MD visualization plugin that:

Use DICOM images already at int.eu.grid SEs orregister new ones (with anonymization)

Would run the brain parameters map creationapplication using TSVD or Tikhonov

Using them would create the core and penumbramap.




Other Future Work

Implementing alternative techniques for getting theperfusion and hemodynamic parameters:

Frequencies filteringParametrical approaches

Bayesian

Comparing all these techniques with simulated and realdataUsing Monte Carlo simulations to check the methods robustnessagainst different levels of SNR

Adapting the application for other medical uses of CTperfusion:

Brain tumors

CT Brain Perfusion Studies

Documents