Next: Bias Field Correction through
Up: Segmentation of Brain MR
Previous: Experiments
Segmentation of Brain MR Images With Bias Field Correction
Since it is a complete approach to segmenting piecewise-constant
images, the HMRF-EM framework can be applied to brain MR images.
However, MR images are often corrupted by a low-frequency
spatially varying artifact known as the bias field. This is
mainly caused by the variation of the interaction between the
human body and the magnetic field. The bias field can also be due
to the differential attenuation of signals and nonlinearities of
the sensitivities of the receiver coils, etc. Although such an
artifact has little impact on visual diagnosis, the performance of
most automatic image analysis techniques, especially
intensity-based segmentation, can degrade dramatically. Therefore,
a robust, automatic, and inexpensive way of correcting for this
artifact is required.
Yongyue Zhang
2000-05-11