The autosomal dominant polycystic kidney disease (ADPKD)
is one of the most common genetic disorders. It is
defined by the growth of multiple cysts in both kidneys.
In approximately half of the cases the disease results in
the need for a kidney transplantation at the age of about
60. Over the course of the disease the growth of the
cysts precedes the deterioration of the kidney function.
The determination of the kidney and cyst growth during
the early stages of the disease by imaging methods is
therefore especially important. Furthermore the analysis
of the cyst growth by imaging methods can give new
insights about the dynamic development of individual
cysts.Segmentation of the kidney:
The first step in measuring the cyst and kidney volume is
the accurate Segmentation of the whole kidney (including
the cysts). The challenging aspects of such a
segmentation are the close proximity of the liver, that
usually also contains many cysts, as well as the
deformations due to the cyst growth. Especially these
deformations make any approach that tries to incorporate
prior knowledge unfeasible.
Accordingly a semi-automatic random-walk algorithm was
chosen for this task. The algorithm is based on a manual
initialization of points inside the object of interest
and derives from these with the help of gradient
information the probability for other pixel to be part of
the object. The advantages of this method are a simple
and intuitive usage, as well as the ability to segment
even weak object boundaries.
Segmentation of the cysts:
The segmentation of the individual cysts is also done in
a semi-automatic manner, based on a watershed
transformation. The cysts can be segmented individually
which makes statistic evaluations concerning the
distribution of the cyst sizes possible.
Additionally simple thresholding is applied in order to
capture very small cysts that usually consist of only a
few pixels. One of the goals here is to determine if
there is a correspondence between the cyst sizes, their
relative frequency and kidney function.