Dockerfile

# scitran/dicom-mr-classifier
#
# Use pyDicom to classify raw DICOM data (zip) from Siemens, GE or Philips.
#
# Example usage:
#   docker run --rm -ti \
#        -v /path/to/dicom/data:/data \
#        scitran/dicom-mr-classifier \
#        /data/input.zip \
#        /data/outprefix
#

FROM ubuntu:trusty

MAINTAINER Michael Perry <lmperry@stanford.edu>

# Install dependencies
RUN apt-get update && apt-get -y install \
    python \
    python-dev \
    python-pip \
    jq \
    wget

# Install scitran.data dependencies
RUN pip install \
  numpy==1.15.1 \
  pydicom==0.9.9 \
  python-dateutil==2.6.0 \
  pytz==2017.2 \
  tzlocal==1.4 \
  nibabel==2.2.1

# Make directory for flywheel spec (v0)
ENV FLYWHEEL /flywheel/v0
WORKDIR ${FLYWHEEL}
COPY run \
     run_classifier \
     manifest.json \
     ${FLYWHEEL}/

# Create Flywheel User
RUN adduser --disabled-password --gecos "Flywheel User" flywheel

# Add code to determine classification from dicom descrip (label)
COPY classification_from_label.py ${FLYWHEEL}/classification_from_label.py
RUN chmod +x ${FLYWHEEL}/run* && chown flywheel ${FLYWHEEL}/classification_from_label.py

# Copy classifier code into place
COPY dicom-mr-classifier.py ${FLYWHEEL}/dicom-mr-classifier.py

# Set the entrypoint
ENTRYPOINT ["/flywheel/v0/run"]