The codes for the measurement of the so-called "patch performance" can be found at: https://github.com/IPMI-ICNS-UKE/patch-performance
The patch performance lets you answer the question "What is the optimal stride/ patch-to-patch overlap/ center crop" for your model and data "on the go", e.g., via training.
As shown in Widening the Focus: Biomedical Image Segmentation Challenges and the Underestimated Role of Patch Sampling and Inference Strategies, this will usually be much more efficient than the theoretical patch overlap or center crop as derived form the model architecture.