You didn't indicate what kind of source you are looking for, If you are trying to localize an iron splinter, you can use a magnetic dipole model. This and biologically generated magnetic fields will be near DC (except for the magnetomyogram). That means that you will have to move the palm relative to the sensors and digitize the palm position in order to have an AC signal that you can localize. Remember, DC SQUID sensors do not have an absolute baseline. That is why some movement (or AC signal) will be needed.

Thanks for the advice. We are studying low frequency biomagnetic fields near DC. We already digitized palm position and the MEG algorithm is also based on a dipole model but is for a spherical object such as a head and not applicable to human palm. Is there any detailed algorithm applicable to human palm? 

You can use the dipole in a homogeneous conducting sphere forward model as an approximation -- if you set the origin far below the center of the palm. This would be equivalent to a dipole in a half-space model. There will be some error due to the return currents being constrained to the palm, which does not extend past you origin. Nonetheless, you should be able to obtain a useful dipole fit.