About the Jacobian abnormality penalization

[yhhuang98]

Hi, thank you so much for this perfect work!

I'm quite curious about the regularization of Jacobian of transformation in your implementation. Seems it penalizes if Jacobian >= 4 or <= 0.2. Commonly we considered those negative Jacobians are unacceptable, but you extend this empirical criterion. May I know if you have any specific consideration here, especially from medical imaging aspects?

Thank you again for this.