Question about the MedXpertQA pathology subset

[HovChen]

Hey there,

Awesome work on Patho-R1! It's really cool that you've made it open-source.

I had a quick question about your paper. In Appendix C.2.1, I saw that for your benchmarks, you "selected 90 pathology cases" from the MedXpertQA dataset.

I was just wondering how you picked those 90 cases. I couldn't find the details in the paper, like if you filtered by certain keywords or had experts manually select them.

Thanks again for the great project!