Code for the paper "A Lightweight Multi-Section CNN for Lung Nodule Classification and Malignancy Estimation" Paper Link
Download the LIDC dataset from the link: https://wiki.cancerimagingarchive.net/display/Public/LIDC-IDRI
Extract the nodule volumes using the PyLIDC tool: https://pylidc.github.io/
Using the extracted nodule volumes extract the cross-sections using the code provided in the notebook to train the model
Code for generating the cross sections is shared in the notebook
Stage 1 training is done by doing transfer learning using all the generated cross-sections.
Stage 2 training is done on the maxpooled features for each cross-section.