Inquiry about Full Supplementary Materials and SA-Med2D-20M Subset Selection

[megumi-qg]

Hi Authors,

Thank you for open-sourcing the MediSee project! I am very interested in your work on Medical Reasoning Segmentation and Detection (MedSD) and the MLMR-SD dataset.

I have two questions regarding the implementation and data details:

1. Full Supplementary Document: I noticed that the tables in the repository screenshots are labeled starting with "D" (e.g., Table D), suggesting there are preceding sections (Appendices A-C). Is there a complete PDF version of the supplementary materials or an official download link available? I searched the ACM Digital Library and arXiv but couldn't locate the supplementary materials .

2. SA-Med2D-20M Subset Selection: In the paper, you mention using a "subset of high-quality images, masks, and labels" from SA-Med2D-20M for both MLMR-SD construction and the mixed training phase. Could you please share the specific criteria or filtering strategy used to define "high-quality" (e.g., based on specific organs, image resolution, or mask completeness)?

Looking forward to your reply!