Transcript A Survey of Medical Image Registration

A Survey of Medical Image
Registration
J.B.Maintz,M.A Viergever
Medical Image Analysis,1998
Medical Image




SPECT (Single Photon Emission
Computed Tomography)
PET (Positron Emission Tomography)
MRI (Magnetic Resonance Image)
CT (Computed Tomography)
Image Modalities
Anatomical
Depicting primarily morphology
(MRI,CT,X-ray)
 Functional
Depicting primarily information on the
metabolism of the underlying anatomy
(SPECT,PET)

Medical Image Integration


Registration
Bring the modalities involved into
spatial alignment
Fusion
Integrated display of the data involved
Matching, Integration,Correlation,…
Registration procedure



Problem statement
Registration paradigm
Optimization procedure
Pillars and criteria are heavily interwined and have
many cross-influences
Classification of Registration
Methods
Nature of
Dimensionality Registration
basis
Nature of
transformation
Domain of
Interaction
transformation
Optimization
procedure
Modalities
involved
Object
Subject
Dimensionality
Spatial dimensions only




2D/2D
2D/3D
3D/3D
Time series(more than two images),
with spatial dimensions




2D/2D
2D/3D
3D/3D
Spatial registration methods


3D/3D registration of two images
2D/2D registration
Less complex by an order of magnitude both where
the number of parameters and the volume of the
data are concerned.

2D/3D registration
Direct alignment of spatial data to projective data, or
the alignment of a single tomographic slice to spatial
data
Registration of time series
Time series of images are required for various reasons
 Monitoring of bone growth in children (long time
interval)
 Monitoring of tumor growth (medium interval)
 Post-operative monitoring of healing (short interval)
 Observing the passing of an injected bolus through a
vessel tree (ultra-short interval)
Two images need to be compared.
Nature of registration basis
Image based


Extrinsic
based on foreign objects introduced into the imaged space

Intrinsic
based on the image information as generated by the patient

Non-image based (calibrated
coordinate systems)
Extrinsic registration methods


Advantage

registration is easy, fast, and can be automated.

no need for complex optimization algorithms.
Disadvantage



Prospective character must be made in the pre-acquisition
phase.
Often invasive character of the marker objects.
Non-invasive markers can be used, but less accurate.
Extrinsic registration methods


Invasive
Stereotactic frame
Fiducials (screw markers)
Non-invasive
Mould,frame,dental adapter,etc
Fiducials (skin markers)
Extrinsic registration methods


The registration transformation is often
restricted to be rigid (translations and
rotations only)
Rigid transformation constraint, and
various practical considerations, use of
extrinsic 3D/3D methods are limited to
brain and orthopedic imaging
Intrinsic registration methods



Landmark based
Segmentation based
Voxel property based
Landmark based registration

Anatomical
salient and accurately locatable points of the
morphology of the visible anatomy, usually identified
by the user

Geometrical
points at the locus of the optimum of some geometric
property,e.g.,local curvature extrema,corners,etc,
generally localized in an automatic fashion.
Landmark based registration

The set of registration points is sparse
---fast optimization procedures

Optimize Measures




Average distance between each landmark
Closest counterpart (Procrustean Metric)
Iterated minimal landmark distances
Algorithm



Iterative closest point (ICP)
Procrustean optimum
Quasi-exhaustive searches, graph matching and dynamic
programming approaches
Segmentation based registration

Rigid model based
Anatomically the same structures(mostly surfaces) are
extracted from both images to be registered, and used as the
sole input for the alignment procedure.

Deformable model based
An extracted structure (also mostly surfaces, and curves) from
one image is elastically deformed to fit the second image.
Rigid model based

“head-hat” method
rely on the segmentation of the skin surface from CT,MR, and
PET images of the head

Chamfer matching
alignment of binary structures by means of a distance transform
Deformable model based

Deformable curves
Snakes, active contours,nets(3D)

Data structure
Local functions, i.e., splines

Deformable model approach
Template model defined in one image
template is deformed to match second image


segmented structure
unsegmented
Voxel property based registration


Operate directly on the image grey
values
Two approaches:


Immediately reduce the image grey value content
to a representative set of scalars and orientations
Use the full image content throughout the
registration process
Principal axes and moments based


Image center of gravity and its principal
orientations (principal axes) are computed
from the image zeroth and first order
moment
Align the center of gravity and the principal
orientations


Principal axes :Easy implementation, no high
accuracy
Moment based: require pre-segmentation
Full image content based


Use all of the available information
throughout the registration process.
Automatic methods presented
Paradigms reported




Cross-correlation
Fourier domain based ..
Minimization of variance
of grey values within
segmentation
Minimization of the
histogram entropy of
difference images




Histogram clustering
and minimization of
histogram dispersion
Maximization of mutual
information
Minimization of the
absolute or squared
intensity differences
…
Non-image based registration
Calibrated coordinate system
 If the imaging coordinate systems of the two
scanners involved are somehow calibrated to
each other, which necessitates the scanners
to be brought in to he same physical location

Registering the position of surgical tools
mounted on a robot arm to images
Nature of Transformation




Rigid
Affine
Projective
Curved
Domain of transformation

Global
Apply to entire image

Local
Subsections have their
own
Rigid case equation

Rigid or affine 3D transformation equation
yi  aij x j
Rotation matrix

r (i ) rotates the image around axis i by an
angle  i
Transformation



Many methods require a pre-registration
(initialization) using a rigid or affine
transformation
Global rigid transformation is used most
frequently in registration applications
Application: Human head
Interaction



Interactive
Semi-automatic
Automatic
Minimal interaction and speed,
accuracy, or robustness
Interaction

Extrinsic methods



Automated
Semi-automatic
Intrinsic methods

Semi-automatic



Anatomical landmark
Segmentation based
Automated


Geometrical landmark
Voxel property based
Optimization procedure
Parameters for registration transformation

Parameters computed

Parameters searched for
Optimization techniques






Powell’s method
Downhill simplex method
Levenberg-Marquardt optimization
Simulated annealing
Genetic methods
Quasi-exhaustive searching
Optimization techniques

Frequent additions:
Multi-resolution and multi-scale approaches

More than one techniques
Fast & coarse one followed by
accurate & slow one
Modalities involved




Monomodal
Multimodal
Modality to model
Patient to modality
Subject



Intrasubject
Intersubject
Atlas
Object

Different areas of the body
Related issues

How to use the registration



Registration & visualization
Registration & segmentation
Validation
Validation of the registration
Accuracy,…