Mesh or NURBS: Which is a better surface input type for patterns design (arm splint)?

May I ask you for some basic advise. I am aiming to achieve something like is depicted in the image below:

Essentially, my overall process is based on a 3D scan geometry reverse engineering (slicing/trimming) processes . I am in a position where I need to decide what type of surfaces I have to use in order to create adjustable and various perforation patterns (Delaunay, Voronoi, Hex, etc.).

Could you please point me what are pros and cons of developing perforations based on a quad mesh input surfaces vs. NURBS input surfaces?

As I understand, the main limitation of a quad mesh surface is that a maximum size of a single perforated unit corresponds to a density (size of a single mesh face) and it can be only changed by decimating an input mesh (loosing accuracy)? On the other hand definitions with surfaces are “heavier” and take more computational time to process. Another thing is the difficulty to smooth edges as can be easily done on thickened quad mesh surfaces with the Catmull Clark subdivision component.

I did see posts like this, but my main concern is to understand what is type of a surface gives you more flexibility in terms of to rapidly generate adjustable organic patterns (maintaining accuracy of the initial input surface)?
Quad mesh or NURBS surface? (19.2 KB)

You might want to see the linked topic. and if you want to get a different density voronoi pattern, this could be a way. (376.1 KB)