Comparing LUNA16 and LUNA2025 Nodule Counts - Clarification Request

Dear LUNA2025 organizers and participants,

I noticed an interesting discrepancy between the datasets and would appreciate clarification to better understand the changes:

LUNA16 contained 1,186 real nodules across 888 scans.

LUNA2025 reports 555 positive nodules and 5,608 negatives.

Key questions:

Why does LUNA2025 have fewer positive nodules despite being a newer challenge? Are these a subset of LUNA16’s annotations, or were they reprocessed with stricter criteria (e.g., size/consensus thresholds)?

Do the 555 nodules in LUNA2025 overlap entirely with LUNA16’s annotations, or were new cases added/removed?

Understanding these differences would help participants contextualize performance comparisons across both challenges. Thank you for your insights!

Caldas, Josafá

Dear Josafa,

LUNA16 and LUNA25 focus on two seperate tasks and utilize different datasets for this.

The LUNA16 challenge focused on the lung nodule detection using the LIDC/IDRI dataset, while LUNA25 focusses on the malignancy risk estimation of lung nodules using annotations in data from the National Lung cancer screening trial (NLST). Hence, the discrepancy between the numbers of both challenges.

Hope to have answered your question.

Kind regards,
Dre Peeters

Dear Dre Peeters,

I appreciate you taking the time to clarify the distinct objectives and datasets used in each challenge—it makes perfect sense.

I’m grateful for your assistance.

Kind regards, Josafa