Statistical tests like the Mann-Whitney test and Kolmogorov-Smirnov test are effective in detecting edges in images by comparing changes in gray-level values between neighboring pixels. These tests work well in both noisy and noise-free images, unlike the T test which is sensitive to noise.