ROCKETSHIP Integration¶

DCEPrep uses ROCKETSHIP (version 1.2+), a MATLAB-based DCE-MRI analysis framework, for two critical steps: T1 map generation and Ktrans pharmacokinetic modeling.

What ROCKETSHIP Does in DCEPrep¶

Pipeline step	ROCKETSHIP function	Script called
Preprocessing Step 10	VFA-based T1 map generation	`run_dce_cli.m`
Analysis Step 1	Ktrans / vp mapping (extended Tofts model)	`parametric_scripts/custom_scripts/T1mapping_fit.m`

How DCEPrep Locates ROCKETSHIP¶

DCEPrep finds the ROCKETSHIP installation at runtime by searching for run_dce_cli.m:

find / -name '*run_dce_cli.m' 2>/dev/null

In Docker, ROCKETSHIP is installed at a fixed path and the script preferences are mounted at:

/opt/ROCKETSHIP/ROCKETSHIP-dev/script_preferences.txt

When running without Docker, ensure the ROCKETSHIP directory containing run_dce_cli.m and the parametric_scripts/ folder is accessible.

Reproducibility¶

DCEPrep records the ROCKETSHIP git commit hash at analysis time and embeds it in case and population reports. This ensures computational reproducibility: you can identify exactly which version of ROCKETSHIP produced a given set of results.

ROCKETSHIP Version¶

DCEPrep requires ROCKETSHIP 1.2 or later with the parametric_scripts add-on.

The GitHub Actions CI/CD workflow automatically clones the petmri/ROCKETSHIP repository (dev branch) before running tests:

- uses: actions/checkout@v4
  with:
    repository: petmri/ROCKETSHIP
    ref: dev

Script Preferences¶

ROCKETSHIP's behavior is configured via a script_preferences.txt file. In Docker, this is provided by DCEPrep's docker/files/ directory and mounted at the expected location.

DCEPrep ships a default script_preferences.txt in docker/files/. Key settings:

Input configuration¶

Setting	Value	Description
`filevolume`	`1`	4D NIfTI input
`dynamic_files`	`/dce/_desc-bfcz_DCE.nii`	Preprocessed DCE series
`t1map_files`	`/anat/space-DCEref_T1map.nii`	T1 map in DCE space
`roi_files`	`/anat/space-DCEref_desc-brain_mask.nii`	Brain mask ROI
`aif_files`	`/dce/desc-AIF_T1map.nii`	AIF T1 map
`start_t`	`3`	Skip first 2 timepoints

Pharmacokinetic model¶

Setting	Value	Description
`quant`	`1`	Quantitative DCE (not semi-quantitative)
`patlak`	`1`	Patlak model enabled (2-parameter, no backflux)
`ex_tofts`	`0`	Extended Tofts disabled
`tofts`	`0`	Standard Tofts disabled
`auc`	`0`	Area-under-curve disabled

AIF and contrast agent¶

Setting	Value	Description
`aif_rr_type`	`aif_roi`	Use provided AIF mask
`aif_type`	`1`	Fitted AIF
`hematocrit`	`0.45`	Default hematocrit
`relaxivity`	`2.8`	r1 relaxivity (mM⁻¹s⁻¹); auto-adjusted based on `AcquisitionDateTime`
`blood_t1`	`2.000`	Blood T1 in seconds
`injection_time`	`-2`	Auto-detect injection time

Processing¶

Setting	Value	Description
`fit_voxels`	`1`	Voxel-wise fitting
`number_cpus`	`0`	Use all available CPU cores
`time_smoothing`	`none`	No temporal smoothing
`xy_smooth_size`	`0`	No spatial smoothing
`outputft`	`1`	NIfTI output

Installing ROCKETSHIP Without Docker¶

git clone https://github.com/petmri/ROCKETSHIP.git
cd ROCKETSHIP
git checkout v1.2   # or the appropriate release tag

Add the ROCKETSHIP directory to MATLAB's path, and ensure the parametric_scripts/ subdirectory is also on the path.

AutoAIF — companion repo providing the neural network for automated AIF detection
GPUfit — CUDA-accelerated curve fitting used by ROCKETSHIP for T1 mapping and Ktrans fitting
ROCKETSHIP on GitHub