SimpleDicomToolkit

Builds a sqlite3 database for all dicom files in a folder. Dicom files can be quickly found by searching dicom fields. Finally (some) images and image volumes can be read directly. SimpleDicomToolkit provides a clean pythonic interface for handling a large collection of dicom files

Usage

build or load database:

db = SimpleDicomToolkit(path='/mydicomfolder')

database have to be build only once. The database is saved to a file and automatically loaded next time. SimpleDicomToolkit will search every time for new folders within the specified path and add them to the database. Removed files from the path will be deleted from the database as well on subsequent loading.

query the database:

db = db.select(PatientName='MyPatient', StudyDescription='MyStudy')

This will select all dicom files that match above specified PatientName and Study Name. Any valid dicom filed can be used to query/select the files.

db.files and db.files_with_fullpath

Will give a list of currently selected files from the database with the relative and fullpath to these files. If no selection is made it will give a list of all files.

db.selection

Will return the currently used filter for selecting the files.

db.reset()

Will remove the current selection filter

db.SeriesDescription

Will give a list of (unique) SeriesDescriptions for the current selection. This works for all dicom fields (PatientName, StudyDescription, etc.)

Reading images

myscript(db.files_with_path)

You can use your own script to read the dicom files by passing the filenames of the current selection.

db.image

In addition after selecting a single dicom series, this may return a SimpleITK a image. It probably works for CT, PET, SPECT, and planar imaging and might work for MRI.

db.array

Will return a numpy array for the given selection.

Advanced usage

db = SimpleDicomToolkit(path='/mydicomfolder', scan=False)

Will load a currently stored database, but will not scan for new files.

db = SimpleDicomToolkit(path='/mydicomfolder', force_rebuild=True)

Will remove existing database and rebuild the database from scratch

db = SimpleDicomToolkit(path='/mydicomfolder', in_memory=True)

Do not create a database file, but only create a temporary database in memory. Database will not be saved.

db = SimpleDicomToolkit(path='/myfolder', SUV=True)

Will convert images to SUV when using db.image and db.array. This works probably for Siemens PET and may or may not work for other vendors due to possible different dicom implementations of SUV values.

db.reset('SeriesDescription')

Will only remove the specified dicom field from the current selection.

Limitations

Small databases up to 10GB should take a couple of minutes to build and can be accessed within seconds after build. Database up to 100GB work quite well, but performance on Windows is much better than linux/macos. sqlite3 is better optimized for Windows it seems. Due to very poor (none at all) database design, databases over 100GB can be very slow to access. SimpleDicomToolkit is primarily intended to be used for relatively small projects.