Biomedical Engineering Reference
In-Depth Information
6
Detecting F
ailure, Assessing Success
J. Michael Fitzpatrick
CONTENTS
6.1 Introduction ................................................................................................ 117
6.2 Measures of Success................................................................................... 118
6.2.1 Alignment Errors ........................................................................... 119
6.2.1.1 Target Registration Error................................................ 120
6.2.1.2 Fiducial Registration Error ............................................ 120
6.2.1.3 Other Error Measures ..................................................... 124
6.2.2 Statistics ........................................................................................... 125
6.3 Methods for Estimating Error .................................................................. 126
6.3.1 Gold Standards............................................................................... 126
6.3.1.1 Computer Simulations ................................................... 127
6.3.1.2 Target Features ................................................................ 128
6.3.1.3 Fiducial Marker Systems................................................ 129
6.3.1.4 Other Standards .............................................................. 130
6.3.2 Registration Circuits ...................................................................... 131
6.4 Accounting for Error in the Standard ..................................................... 132
6.5 Independent Validation............................................................................. 133
6.6 Conclusions................................................................................................. 134
Acknowledgments .............................................................................................. 136
References ............................................................................................. 136
6.1
Introduction
The goal of image registration is to find a geometrical transformation that
aligns points in one view of an object with corresponding points in another
view of an object. In medical applications one of these views will typically be
a tomographic image such as CT, MR, SPECT, or PET, but may include x-ray
or even video images. The other view may be chosen from one of these
0-8493-0064-9/01/$0.00+$.50
© 2000 by CRC Press LLC
117
