Our team determined that the optimum statistic for comparing and evaluating two images is the SSIM. So is there another ideal criteria for comparing and evaluating the quality of images?
SSIM is more than fine for your tests 
MSE would have worked too I guess.