Hello,
I am using the Digitize GUI with polhemus patriot on matlab 2017a and I am having serious problems with the accuracy.
No matter what I do, I always get very imprecise measurements. I followed the tutorial but even when I am trying to digitize the same point without moving the stylus, I still get differences that can be in the order of cm from one measurement to the other. I never made it beyond the digitization of the fiducials because the difference between the sets is > 5mm
I do not think my polhemus is broken: when I use the PiMgr app that comes with polhemus I do get reliable measurements of X,Y,Z.
Azimuth, elevation and roll are a bit less reliable: for those coordinates there can be a variation of about 10 degrees.
Has anyone experienced something similar and has suggestions on how to obtain reliable measurements?
Hi Eva,
The polhemus devices are very sensitive to external interference. Be sure the subject (or test device) is placed away from walls and other metal objects. You may find different reliability at different locations in the room. You may also find different reliability if you move the receiver to different locations with respect to the transmitter. A bit of trial an error may be necessary.
I can also suggest you try to look at the raw points coming from the device. This might help determine the source of the problem. Try this by collecting points at one location and then along a ruler so you have actual distances. You can try the attached code using matlab to grab the raw points.grab_points_polhemus (1.1 KB)
Let me know if you need help with that.
Hi Beth,
I have been not working on this project since a long time and now I am back to polhemus… and the problems that come with it!
I have used the script you sent me and the system seems accurate when measuring on a fixed point (very low variability). I also tried to measure along a ruler and this seems also fine. However, it is really impossible to get decent measurements when trying it on a real head. The cartesian coordinates are always pretty decent however it is quite hard to make reliable measurements of elevation, pitch and roll.
Since brainstorm accounts for them, we never get measurements of Nasion LPA and RPA colse enough with the acquisition of the points on the rest of the head. We were wondering if elevation, pitch and roll are really necessary or we could just record x,y and z and use those coordinates, since brainstorm seems to reduce everything to 3 coordinates anyway.
I apologize in advance if the question is a bit naive.
