Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 1

The template-less or typeless layer concept's goal is to hide ITK's template parameters

from the user. A common suggestion was to simply consider all images as 32-bit or 64-
bit real numbers. However, this approach can cause unacceptable memory use and
performance penalties. Consider starting with an image that is an 8-bit and 4 gigabytes.
Converting to a 64-bit type would use eight times the memory now requiring 32
gigabytes of memory for its representation. This monolithic choice changes the nature of
working with an image from one that can be done on a laptop to one that requires a
large server. Additionally, the users surveyed demanded the flexibility to use a variety of
pixel types.

Therefore the template-less layer concept requires that SimpleITK images all have the
same external type and interface, unlike ITK. The details of dealing with the different
ITK template parameters is hidden so that the user can generically manipulate any
image. An image's pixel type and dimension are handled by SimpleITK internally and
not directly exposed to the user, unlike ITK which requires the user to explicitly
determine them at compile-time. At runtime a PixelIDValue, in the form of an integer
and enumerated type is used to represent a pixel type, which along with the image
dimension are intrinsic run-time attributes of the SimpleITK image.

You might also like