When the deep learning method is applied to the diagnosis of benign and malignant CT images of lung nodules, what are the proprietary (unique) data characteristics that can be used/concerned?

Deep Learning
1

Personally believe that effective, full use / attention to the characteristics of the data set, in order to better play the methodology. But it is not that the method itself (deep learning model) is not important. The unique features of the data set and the model built should be complementary.

2

Are you referring to 3D data from a CT or 2D slices from a CT scan?