This preview shows pages 1–2. Sign up to view the full content.
Summary Report – ModelBased Estimation for Dynamic Cardiac
Studies Using ECT
Group Member: Wan Huang, Yiying Zhu, Yue Hou
I.INTRODUCTION
In this report, modelbased estimation using Emission Computed Tomography (ECT) for dynamic
cardiac study is discussed about, which has multitude of advantages over existing imaging techniques.
Through imaging an organ of interest over time using ECT, one can observe the dynamic behaviour
of the radiotracer, thus quantify the cardiac processes.
The estimation process can be modelled as: Using measurements of plasma tracer concentration as
input and tomographically measuring the organ tracer concentration as output, the object is to determine
the compartmental parameters as the intermediary system. [1]
Conventionally, human operators are needed to specify myocardial regions of interest (ROI) for each
image and function of detected counts versus time index is calculated. Estimation algorithm is only used
then to underlie the concentration change in each ROI from the counttime function. [1]
Two factors obstruct accurate qualification of ROI when ECT is applied, which are limited system
resolution and measurement noise, [two difficulties] where limited system resolution causes systematic
error (called bias) because adjacent regions would contaminate each other thus the boundary is difficult to
define.
Existing researches are divided to two directions: Postreconstruction correction and ROIbased
estimator based on maximum likelihood (ML) criteria. [2]
For ROIbased ML estimators, [2] shows statistically unbiased ROI concentration estimates can be
generated if ROIs are specified exactly. He formulated parametric image model as following:
(1)
The formula allows estimation of compartmental parameters (X) directly from the projection data (Y)
and system response matrix accounting for resolution (W).
Both postreconstruction correction and ROIbased estimators assume perfect ROI delineation from
reconstructed ECT images [1], which is never true. Consequently, erroneous ROI specification leads to an
extra source of error.
The research paper is to establish a strategy that jointly estimating boundaries as well as
compartmental parameters.
II.
OBSERVATION MODEL
The author proposed an observation model to parameterize compartmental parameters, myocardial
boundaries, left ventricular input function and background concentration. [1] Based on this, estimator is
applied. The model has a more generalized form than [2] as:
(2)
Where Ψ is degradation factor that depends on boundary S, and C is ROIbased concentration vector
that depends on compartmental parameter Θ.
Two assumptions are made for simplification: 1. On short axis section, the object consists of three
homogeneous regions: left ventricle, myocardium, and background. 2. The heart is stationary.
A simple observation model can be described as:
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
This is the end of the preview. Sign up
to
access the rest of the document.
This note was uploaded on 01/10/2012 for the course EECS 551 taught by Professor Wakefield during the Spring '08 term at University of Michigan.
 Spring '08
 Wakefield

Click to edit the document details