luna16

LUNA16 Lung Nodule Analysis

Respository containing code to port ZNet Solution to CAFFE frameowrk.

A rough list of requirements:

Python 2.7 with tqdm, pandas, numpy, scipy, scikit-image, scikit-learn, opencv2 and CAFFE.

UNET for dense prediction

Preprocessing

Convert data to 1x1x1mm_512x512 slices. A requirement is also a set of segmentations of the lungs (can be found on the LUNA16 website). Place your data in folder data/original_lungs:

Use script src/data_processing/create_same_spacing_data_NODULE.py, this may take a long time (up to 24 hours) depending on your machine.

Then, download imagename_zspacing.csv from here and put it in the data folder.

Unet training

src/deep/unet/caffe_unet_train.sh

Parameter configurations is in params.py

Unet held out set dense predictions

src/deep/caffe_unet_predict.sh

When training a model a folder is created in /snapshots/unet. The folder name is the model name here. Manually look up which epoch had the lowest validation set loss and was a checkpoint.

Unet dense predictions -> candidates:

cp -rf ./snapshots/unet_iter_311061.caffemodel_predictions/* ./results
python candidates.py
python candidate_merging.py

a finalizedcandidates_unet_ALL.csv will be generated in data folder

False positive reduction (Wide ResNet)

Preprocessing:

cp ./data/finalizedcandidates_unet_ALL.csv ./csv
. ./src/data_processing/create_wrn_input.sh

Training

. ./src/deep/resnet/caffe_resnet_train.sh

Predict

. ./src/deep/resnet/caffe_resnet_predict.sh

The prediction CSV can then be found in the model folder. All you have to do now is combine these. You could use src/ensembleSubmissions.py for this, which also features some equalization of predictions of the different models.