In the project "lung_nodule_ct_detection", how should I determine the size of the nodule?

[dqy818]

Each nodule has three dimensions: length, width, and height. In the LUNA16_datasplit JSON file, I noticed that the nodule sizes are the same across the three dimensions, but in reality, they are different. Which dimension should I use as the size for training data—the maximum, or the minimum of the three dimensions?

[yiheng-wang-nv]

Hi @Can-Zhao , could you help to provide some suggestions here? Thanks!

[Can-Zhao]

This is because this training data only provides the radius and center of each nodule. If you have a more accurate nodule annotation, you can let the three dimensions be different.