DVH Analytics is a software application to help radiation oncology departments build an in-house database of treatment planning data for the purpose of historical comparisons and statistical analysis. This code is still in development. Please contact the developer if you are interested in testing or collaborating.
The application builds a SQL database of DVHs and various planning parameters from DICOM files (i.e., Plan, Structure, Dose). Since the data is extracted directly from DICOM files, we intend to accommodate an array of treatment planning system vendors.
In addition to viewing DVH data, this software provides methods to:
- download queried data
- view plan contours
- create time-series plots of various planning and dosimetric variables
- calculate correlations
- and generate multi-variable linear regressions.
The code is built upon these core libraries:
- pydicom - Read, modify and write DICOM files with python code
- dicompyler-core - Extensible radiation therapy research platform and viewer for DICOM and DICOM RT
- Bokeh - Interactive Web Plotting for Python
For installation instructions of the source code, see our installation notes.
We're working on an easier way to deploy DVH Analytics here.
Some installation challenges with DVH Analytics that are resolved with Docker:
- Complete DVH Analytics docker image is used.
- A workaround for this Bokeh bug is applied.
- Postgres SQL is included (so no need for setting up a database and user access).
- All three servers for the main, admin, and settings views are started.
DOI: https://doi.org/10.1002/acm2.12401
“DVH Analytics: A DVH Database for Clinicians and Researchers,” J. App. Clin. Med. Phys. - JACMP-2018-01083
- Python 2.7
- Python 3 not currently supported
- Future 0.16.0
- for bridging gap between 2.7 and 3.x
- we're still working on the Python 3 transition for DVH Analytics
- PostgreSQL and psycopg2
- SciPy
- pydicom 0.9.9
- shapely 1.6b2
- Statsmodels 0.8.0
- dicompyler-core 0.5.3
- requirements per developer
- numpy 1.2 or higher
- pydicom 0.9.9 or higher
- pydicom 1.0 is preferred and can be installed via pip using: pip install https://github.com/darcymason/pydicom/archive/master.zip
- matplotlib 1.3.0 or higher (for DVH calculation)
- six 1.5 or higher
- requirements per developer
- Bokeh 0.13.0