SEMINARIESCHEMA FÖR MATEMATISK STATISTIK 

Tisdag     6/8  13.15         Fred Godtliebsen
          1996                Department of Mathematics and Statistics,
                              University of Tromso:

                              Statistical Methods Applied in Medical Imaging



Abstract:

In medical applications information is frequently presented as two- or
three-dimensional images.  Unfortunately, such images are typically
disturbed by different kinds of distortions.  This talk will
concentrate on magnetic resonance (MR) images although most of the
methods can be applied in other areas as well.  The (main) distortion in MR
images can be characterized as independent and identically distributed
normal noise.  Three different aspects will be discussed:

1)  Noise reduction.
2)  Estimation of underlying tissue types.
3)  Dynamic MR images.

In part 1) we study local constant (and local linear) M estimators
empirically and theoretically.  Part 2) is solved by applying density
estimation to binned data.  From a practical point of view, the most
important part of this talk is the description of statistical methods
applied to dynamic MR images.
Combining parts 1) and 3) makes the information available in dynamic
MR images easier to interpret for radiologists.




Lokal: Rum 227 i Mattehuset.

Björn Holmquist 046-2228546