Leave-One-Out Cross-Validation for small dataset

[Meddebma]

Dear MONAI,

I am trying to classify two different abdominal diseases in a small dataset (n=120), and the accuracy with DenseNet ist about 0.8, which is really good. Due to the small size I only made a training and a test dataset (0.8/0.2). I've been reading that Leave-One-Out Cross-Validation is pretty good for small datasets. Do you have a module how to implement LOOCV during training, or maybe give me some hints how to do it?

here is the notebook:
https://colab.research.google.com/gist/Meddebma/6bb6c4ed7458a76dc56efb74b3e5a00f/3d-classification.ipynb#scrollTo=Nm0LXsvlgp_N

Thank you very much!