Patch2Self_MATLAB

Unofficial re-implementation of Patch2Self in MATLAB

Unofficial Re-implementation of Patch2Self in MATLAB

• Only support OLS and Ridge regression

  • Internally solves the corresponding normal equations using mldivide
  • For OLS: beta = (X.' * X) \ (X.' * y); i.e., <math-renderer class="js-inline-math" style="display: inline" data-static-url="https://github.githubassets.com/static" data-run-id="302c3460819fce4a8c96515b54de3b1c">$$\min_\beta (X^TX)^{-1}X^Ty$$</math-renderer>
  • For Ridge regression: beta = (X.' * X + alpha * eye(size(X, 2))) \ (X.' * y); i.e., <math-renderer class="js-inline-math" style="display: inline" data-static-url="https://github.githubassets.com/static" data-run-id="302c3460819fce4a8c96515b54de3b1c">$$\min_\beta (X^TX + \alpha I)^{-1}X^Ty$$</math-renderer>

• Does not support clip_negative_vals, shift_intensity, verbose (yet)

• Caution: I haven't thoroughly tested the code. Feel free to try it out and let me know if the performance is equivalent to that of the official implementation.

References

[Fadnavis20] S. Fadnavis, J. Batson, E. Garyfallidis, Patch2Self: Denoising Diffusion MRI with Self-supervised Learning, Advances in Neural Information Processing Systems 33 (2020)

Adopted from

Patch2Self (DIPY)