@Arcana @georgia would not say "impossible" of such; just very difficult.
to begin with, you can't just inject stem cells into some tissue and expect them to properly integrate. human bodies grow stepwise all together all at once, so a cell needs all sorts of signals incoming to tell it what to do, which usually happen at specific times during development. some things are easier than others—e.g. you can kinda replenish hair follicles—but there are still issues with getting them to integrate properly. they'll be oriented wrong and you'll have hair grow out sideways. or, as in current experiments with cardiac muscle currently, it'll look ok but then beat arrhythmically. these level problems seem difficult, much more so than early researchers anticipated, but still might be tackled with simple chemical signalling or something. the bigger issues though are
- cancer and general mutative degradation, which happen everywhere diffusely and you can make worse trying to replace tissues this way, bumping them up the differentiation tree (not really a tree; much fuzzier)
- brains, the tissue you care most about for this problem and also the most difficult one to approach. for tissues generally there's the problem of getting them properly vascularised and ennervated ete (how?), and for a brain you can't even try some print-and-transplant method; it all has to happen in situ in the most complicated tissue you have, with all these subtly differentiated cells grown into an ad-hoc pattern and with long-range connections strung between them. drop new cells in here and it's easy to give yourself a tumour, but actually integrating and not just messily disrupting function of your ffa or something is a "who even knows" kind of problem
toss in several other side problems, like purging bioaccumulated toxins, and if it's ever to be done at all still regenerating so is nowhere near yet feasible and won't be for a long time