Biomedical Engineering Reference
In-Depth Information
or color. To achieve useful displays, researchers have used arrow pic-
tures in which small variable length arrows are scattered over the display
plane, with the direction of each showing the local direction of shift, and the
length indicating the magnitude of the shift. Other methods include place-
ment of a regular grid on the original image and displaying the distorted
grid after registration is complete (see Chapter 13). Color coding of displace-
ment has also been tried. Techniques for displaying deformation informa-
tion are likely to develop further as applications for nonrigid registration are
explored and precise requirements for the information to be extracted are
defined.
and
4.7
Conclusion
Marshaling the data required to register images can entail a number of steps,
ranging from file format conversion to image preparation by intensity correc-
tion, image segmentation, etc. Careful data preparation is essential to avoid
errors that result in failed registration or data that are at risk of incorrect inter-
pretation. However, the steps required are conceptually simple and, once
established, can be made automatic or at least quick and efficient to operate.
Such data preparation is likely to be required even when ideal scanner per-
formance is assumed. In practice it is often necessary also to take account of
imperfections in the data acquisition process itself. Such imperfections may
cause errors that it is essential to correct or at least be aware of. The nature
and consequences of these errors are the subject of Chapter 5.
Once image registration becomes a core part of a medical imaging application,
it has consequences that have an impact on the whole imaging process, from the
way examinations are performed to data acquisition methods, to image han-
dling, image analysis, and, finally, on image presentation and interpretation.
Some of the factors discussed in this chapter have implications for data acqui-
sition efficiency, archiving, and processing strategies. It is the power of image
registration to enhance and add value to medical imaging that makes these
costs worth accommodating.
References
1.
Noll, D.C., Boada, F.E., Eddy, W.F., A spectral approach to analysing slice
selection in planar imaging: optimisation for through-plane interpolation.
Magn. Reson. Med.,
38: pp. 151-160, 1997.
2.
Stark, D.D. and Bradley, W.G. (eds.),
Magnetic Resonance Imaging
third ed.,
Mosby, St. Louis, 1999, Chapter 4.
3.
Stark, D.D. and Bradley, W.G. (eds.),
Magnetic Resonance Imaging
third ed.,
Mosby, St. Louis, 1999, Chapter 5.
