edge effects in butterworth filter and hilbert transform

Dear experts,

I am filtering an fMRI signal using a 7th order Butterworth filter (0.04-0.07 Hz), and after that I extract the phase using the Hilbert transform. I have read about edge effects in both methods. I was wondering what your take is on dealing with these edge effects.

I have 395 datapoints, sampled at 1 every 1.6 sec. I was thinking of removing datapoints at the beginning and end of the signal so that at least one cycle of the lowest frequency has passed (1/0.04 = 25 / 1.6 = 16 datapoints) to stay on the safe side. Does this seem reasonable to you?

Thanks very much for you input.

Best

Hans