Pardi

A command line tool to catalog DICOM files. For every file it finds, it extracts a configurable set of DICOM fields (by default Patient ID and Patient Name) plus the file path.

Build

To build the cli tool, move into the pardi folder and run cargo build --release. This command generates an optimized build in the target/release folder

To build the tool with rayon enabled, run cargo build --release --features rayon

Usage

After the build process, in target/release you can find the pardi binary. Run it with ./pardi. See the next sections for the available options.

Options

• --path = Specify the path where to look for DICOM files. By default it tries to scan the current directory

Example: ./target/release/pardi --path /home/user/data

• --format = The format of the output, json and csv are available. Default json

Example: ./target/release/pardi --path data --format csv

• --debug = Print error messages for unprocessable files. Default false.

Example: ./target/release/pardi --path data --debug

• --output = Save the catalog to a file. If no output option is speficied, it prints the catalogue on screen.

Example: ./target/release/pardi --path data --format json --output catalog.json

• --tag = A DICOM field to extract, given by its standard dictionary alias (e.g. PatientID, PatientName, StudyDate, Modality, SeriesInstanceUID...). Repeatable. If omitted, defaults to PatientID and PatientName. The requested field names become the CSV column headers / JSON object keys, in the order given.

Example, extracting patient ID together with the modality and study date of each file: ./target/release/pardi --path data --format csv --tag PatientID --tag Modality --tag StudyDate

Any alias known to the DICOM standard data dictionary works; see the dicom-dictionary-std documentation for the full list, or inspect a sample file with a DICOM dump tool to see which attributes it carries. If a requested field is missing from a given file, that file is skipped and reported as an error (visible with --debug).

Benchmarks

The project has a Criterion benchmark suite in benches/catalog_benchmark.rs covering DICOM file parsing.

Run it with:

cargo bench

HTML reports (with plots) are written to target/criterion/report/index.html.