Edit: perhaps a set of crowd-sourced, extremely explicit instructions about how to manufacture, purify, and test such antibodies would be a start. With such a set of instructions, they might be able to find experts to help them under the table.
Even theft isn't an option here, unless you're somehow certain that a particular lab has exactly the right antibody. (In my lab, we often have trouble keeping track of that for ourselves...)
And don't forget that you need to be able to maintain sterile conditions, and have whatever other reagents you need to dilute things, etc.
http://www.biolegend.com/percp-cy55-anti-human-cd47-4338.htm...
Of course, to get CD47 depletion, you'd probably want an antibody that isn't conjugated (functional grade):
http://www.ebioscience.com/resources/in-vivo-use-of-antibodi...
Also, some more help on the terminology: CD47 (and most, if not all molecules that fall under the Cluster of Differention [CD] nomenclature) vary on a per-species basis. So an antibody that targets CD47 in mice is not going to stick to human CD47. "Anti-" in this context means that the antibody in question will stick to specific molecule in the species that comes after the "anti-".
Note what it says on the bottom of the page. That's there because of regulations.
In brief: you will hurt people doing this and achieve nothing.
Morally speaking, the judgement of a patient (and the physician he has chosen) must not be impeded even for an instant. The FDA renders individual judgement on these matters irrelevant, and the result is often death and slowed medical progress.