Individualized Region Detection with Total Variation

Citations

Wu, S., Wang F. and Feng, L. (2023) Individualized Region Detection with Total Variation.

Environment and usage

The following environments are required:

• Python 3.9.5 (anaconda is preferable)
• Pytorch '1.10.2+cu113'

Code

We provide two examples: real MRI images with simulated signals and MNIST-bags.

The detailed usage refer to the Tumor2d.ipynb and MNIST.ipynb

Real images with simulated signals

An illustration of settings and our model performance. The red rectangles indicate the detected regions with top-5 probability within the local matrix of the image.

MNIST-bags

An example illustration of our model prediction for a positive collection. The green line at the bottom indicates the patch with the highest probability among the top-5 patches.