Source code for the CVPR 2015 and JMLR papers:

• Interleaved Text/Image Deep Mining on a Very Large-Scale Radiology Database (link)
• Interleaved Text/Image Deep Mining on a Very Large-Scale Radiology Database for Automated Image Interpretation (arxiv-link)

Required software packages:

• Caffe - release candidate 2 or older
• Java or Scala - to run the Stanford Topic Modeling Toolbox (link)
• Python 2.7 with the following packages:
  • Numpy
  • Scipy
  • gensim
  • h5py
  • Matplotlib
  • Scikit-Learn
  • Scikit-Image
  • NLTK