A. F. Lukban, G. W. Dean, R. Lisbona, F. Dubeau,
D. McMackin and
A. C. Evans
1 - INTRODUCTION:
The use of single photon emission computed tomography (SPECT) in the
investigation of cerebral disorders such as epilepsy has generally
been limited to its utilisation as an aid in the gross lateralisation
of the epileptogenic focus [3][2][1]. The
combination of SPECT's continuously improving resolution capabilities
and very attractive cost benefit ratio as compared to positron
emission tomography (PET) has served to increase its use in the
investigation of more finely localised functional studies.
Compared to PET, the advent
of more finely localised investigations with SPECT was simply not
possible with the resolution capabilities of systems from just a
decade ago. Due to coarse resolution and the functional nature of the
information provided by SPECT scans, such studies require the anatomical
information and very fine resolution provided by co-registered magnetic
resonance (MR) images.
The internal landmark matching (ILM) technique was first implemented for the registration of multi-modality brain volumes by Evans et al [5][4] to provide for such procedures. The transformation errors arising from the use of the ILM technique for registration were investigated by using point simulations and real scan studies based on a 3-D brain phantom with external fiducials. The results of previous validations of the use of the ILM technique in PET/MR registrations may not be extended directly to SPECT/MR registrations because of the anisotropic nature of SPECT's spatial resolution and differences in the methods used to measure error. The use of SPECT/MR registration with the ILM technique for effective anatomical localisation of regional cerebral perfusion in the study of epilepsy is investigated.
2 - METHOD:
The ILM technique incorporates user chosen homologous point pairs from
the MR and SPECT volumes in the Procrustes algorithm for the
anatomical/functional volumetric registrations. The Procrustes
algorithm finds the best fit tansformation in the matching of the
images by minimising the mean squared distances between corresponding
point pairs. The errors in registration were investigated with point
simulations and the Hoffman brain phantom fitted with external
fiducials to provide a standard for true registrations. Point
simulation results with spheres of different radii, homology error, and
number of point pairs were compared to results from a real scan with
the brain phantom to allow a more effective study of the translation
and rotation errors which arise in transformations required for
clinical registrations (see figure 1).
3 - RESULTS &DISCUSSION:
From point simulations on SGI workstations,
the standard deviation,
, of the translation errors were
constant with different radii of spherical boundaries of the
dispersion of points and generally
increased with increase in homology error (approximately the
magnitude of the resolution of the functional image) and decrease in the
number of point pairs used (see figure 2).
of
the rotation errors behaved in the same way with respect to homology
error and the number of point pairs, but decreased with increased
radii of the spherical boundary of the dispersion of the
chosen homologous point pairs (see figure 3). The real scan
registration errors from the brain phantom corresponded well to those
predicted by the point simulations. For a 25 point registration,
it was found that:
These values represent which, from an underlying
Gaussian distribution of errors, corresponds to a probability of
99.7 %.
With this level of registration accuracy, hippocampal activity
on Tc-HMPAO SPECT images has been investigated with MR based volumes of
interest (VOI). These VOI's are drawn to cover the entire
volume of the hippocampus (from amygdala to fornix).
Two example cases which show the diagnostic enhancement
provided by registered volumes are given in figures 4 and 5.
4 - CONCLUSIONS:
Very good registration accuracy may be achieved with the ILM
technique. The use of registered SPECT and MR volumes is a
useful tool which aids the effective anatomical localisation of
abnormal perfusion on SPECT images. The MR and corresponding
interictal SPECT brain volumes from about 10 patients have been
registered with this method. MR based VOI
analyses have been peformed to measure left-right rCBF asymmetries on the
registered SPECT scans and have greatly aided in the creation of more
objective methods of analyses.