.
The empirical mean and variance of this index are:With 24 images of the same patient, we have only 23 independant
registrations. To obtain a more significant result, we have used 60 randomly
chosen pairs of images and thus obtained 60 values
of the validation index that should be theoretically distributed as .
The empirical mean and variance of this index are:
|
|
|
The comparison of our 60 samples
with the theoretical
distribution using the Kolmogorov-Smirnov test gives:
KS test accepted (significance: 0.59)
which means that our accuracy estimation is fully validated.
[ Back to uncertainty
in registration for medical imaging ] 
