Automatized Methods for Coregistration of PET and MR Brain Images Jonny Reichwald

Automatized Methods for Coregistration of PET and MR Brain

Images

Jonny Reichwald

A: axon, B: neurotransmitter 1,C: neurotransmitter 2,

D: synaptic cleft, E: synapse for neurotransmitter 2, F: dendrite.

The Synapse

Radioisotopes

Isotope Name Half-life (s.) [11C] Carbon-11 1223.0 [15O] Oxygen-15 122.24 [13N] Nitrogen-13 597.9 [18F] Flourine-18 6586.2

A: a radioactive decay occurs which emits a positron. B: the positron collides with an electron and emits photons. C: The photons are detected by photomultipliers

A: The decay emits photon pairs at ~ 180 degreesB: The ideal case of a decay resulting in the LOR D.

PET image after reconstruction

MR image

Production of helmet

PET scan

●Resolution●Noise●Reconstruction●Coregistration●Modelling●Ligands

Limiting factors

●Resolution●Noise●Reconstruction●Coregistration●Modelling●Ligands

Limiting factors

Coregistration

●Finding transform●Coregistration algorithms

●Transform●Resampling

Rigid Body

Affine

Projective

Curved

Transform Models

Coregistration

Coregistration

● Manual– Landmarks

● Intrinsic● Extrinsic

– Image overlay● Automatic

– AIR/SPM99● With automatic skull-stripping

– Mutual Information

Voxel probability

Entropy

Kullback-Leibler distance

Mutual Information

PET-PET Registration

Before After

HistogramPET-PET

Before After

Interpolation

● Nearest Neighbour● Linear● (Windowed) Sinc● Splines

General Interpolation

Nearest Neighbour

Linear

Sinc

Splines

Splines

Splines

Time Activity Curve

TAC error

AC-PC Alignment

AC-PC Alignment

Program Platform

● Hba Irix● ECAT6 Solaris● ECAT7 Solaris● ImageJ Java

compatible● SPM Matlab

compatible● PMOD Java compatible

Conclusion

● Use splines or NN for interpolation● AC-PC alignment not important● SPM● PMOD