The given examples for taking measurements are good, but IMO still offers a certain range for issues. A photo based measurement (via upload, webcam - T-Pose) with put on capture suite may help to get a more accurate measurement and better calibration. Then it should be enough to measure e.g. one main part where you can´t make major mistakes like e.g. total size.
I would also like to know how the position of the neurons affects to a given custom size if put on a little bit different every time.
Would this produce very different results e.g. if the position variies 1-2 cm in 2 directions?