juliandewit/kaggle_ndsb2017

Implements an ensemble lung cancer detection pipeline combining 3D convolutional networks for nodule detection/malignancy prediction with U-Net mass segmentation, trained on LUNA16 and manual NDSB annotations, then blended via XGBoost meta-learning. Uses Keras/TensorFlow with preprocessing that normalizes CT scans to 1×1×1mm voxels and generates lung segmentation masks for efficient inference across multi-fold models.

626 stars. No commits in the last 6 months.