Intensity-based classification of MR images has proven problematic, ev
en when advanced techniques are used, Intrascan and interscan intensit
y inhomogeneities are a common source of difficulty, While reported me
thods have had some success in correcting intrascan inhomogeneities, s
uch methods require supervision for the individual scan, This paper de
scribes a new method called adaptive segmentation that uses knowledge
of tissue intensity properties and intensity inhomogeneities to correc
t and segment MR images, Use of the expectation-maximization (EM) algo
rithm leads to a method that allows for more accurate segmentation of
tissue types as well as better visualization of magnetic resonance ima
ging (MRI) data, that has proven to be effective in a study that inclu
des more than 1000 brain scans, Implementation and results are describ
ed for segmenting the brain in the following types of images: axial (d
ual-echo spin-echo), coronal [three dimensional Fourier transform (3-D
FT) gradient-echo T1-weighted] all using a conventional head coil, and
a sagittal section acquired using a surface coil, The accuracy of ada
ptive segmentation was found to be comparable with manual segmentation
, and closer to manual segmentation than supervised multivariant class
ification while segmenting gray and white matter.