calculate CSA_mean_diff

[jcohenadad]

instead of using CSA variable:

csa-atrophy/csa_rescale_stat.py

Line 361 in f4f5c4b

sample_size_long_80.append(np.ceil((((1.96 + 0.84) ** 2) * (std_diff)) / ((((100 - rescale_area) / 100) * CSA) ** 2)))