Summary:
The paper describes a method to apply ITK filters to slices, i.e. sub-volumes which are one dimension less than the original volume. It processes all slices of the original volume, pipes them through a set of filters and stores the result in an output volume.

Hypothesis:
"Non Applicable".

Evidence:
The method itself does not require any evidence; the statement that "filtering an image in N-1 dimensions, slice by slice, can be very useful in many cases" is hard to prove with numbers but I totally agree with the author  and many examples exist.

Open Science:
The source code as well as samples are provided.

Reproducibility:
I have compiled and used the class extensively.

Use of Open Source Software:
The author's filter is properly derived from ITK's ImageToImageFilter and thus integrates seamlessly into the ITK framework.

Open Source Contributions:
The source code is fully provided. The filter was running out of the box besides the issue discussed under "Additional Comments".

Code Quality:
The code is easy to read and follows the ITK coding guidelines.

Applicability to other problems:
n/a

Suggestions for future work:
In a future version it would be great if a connected set of sub slices could be processed, see also "Additional Comments".

Requests for additional information from authors:
n/a

Additional Comments:
The current revision is always setting the requested region of the output to the largest possible region. I would be nice if full sub-slices would be allowed. I.e. slices of dimension N-1 where

• the first N-1 values of the requested region's index of the output should be equal to zero and
• the N-th value of the requested region's output would be the starting index.

Furthermore

• the first N-1 values of the requested region's size should be equal to the first N-1 values of the largest possible region's size while
• the N-th value of the requested region's size would be actually the number of slices to be processed.

As an example this would allow a user to process single 2D images as well as a set of continuous 2D images while ensuring that full slices are processed which would be intuitively conforming with the class's name.

In the loop in which is iterating over the slices of a volume the upper bound is not computed correctly. The upper bound should be equal to:

requestedSize[m_Dimension] + requestedIndex[m_Dimension] (+, not -).

Besides these remarks a very nice tool!