PRIME196

This is an example pipeline for the radnet project. It is meant to contain:

• bruker_OE_IR_VFA_DCE_session.json -- a session-template that is used to identify sets of files produced by an MRI scanner as belonging to this particular study. See PreclinicalMRI/Bruker_pre_process.py
• OE_IR_DCE_VFA.cwl -- a CWL-workflow that defines the study pipeline, using tool-wrappers (and subworkflows from cwl_madym)
• OE_IR_DCE_VFA_input.yml -- an input-template that can be used to link the files matched by the session-template to the CWL-workflow.

The CWL-workflow and the input-template serve as public references of the study's methodology.

FUTURE: particular instances of this pipeline (together with data) should become Research Object Crates?

OE-IR-DCE-VFA pipeline

Two independent chains: OE-IR, and DCE-VFA, merged at the end only to generate combined-masks.

OE-IR calculates T1 map using Inversion Recovery with Efficiency weighting, before (baseline) and after the delivery of supplemental oxygen (enhancing).

"Elevated concentrations of dissolved oxygen increase tissue's longitudinal relaxation rate (R1). Following the delivery of supplemental oxygen, initially well-oxygenated regions where haemoglobin is well saturated develop an increase in dissolved molecular oxygen with consequential reduction in T1 times." (McCabe et al. 2023)

DCE-VFA (Dynamic Contrast Enhanced MRI) attempts to fit a mathematical model of how a tracer (CA) perfuses through tissue vasculature.

"Quantitative DCE-MRI data acquisitions require three measurements: 1) recording of a map of the native T1 values before contrast administration; 2) acquisitions of T1-weighted images following CA introduction at a reasonably high temporal resolution to be able to characterize the kinetics of the CA entry and exit into tissue; and 3) a method to estimate the time rate of change of the concentration of the CA in the blood plasma, the so-called arterial input function (AIF).
...
The most commonly used method of mapping T1 before administering the CA is to use a series of so-called spoiled gradient-recalled echo (SPGRE) images acquired with different degrees of T1-weighting... The idea is to collect SPGRE data at multiple flip angles [...] and fit the data to Eq. (3) with S0 and T1 as floating parameters." (Yankeelov & Gore, 2009).

A1. OE IR T1-mapping (madym_T1 --T1_method IR_E)

Inputs:

• IR/OE_*.nii.gz volumes (with *.json metadata)
• Parameters: T1_noise, TR (read from metadata)

Outputs:

• Maps (IRE/*.nii.gz) for T1, M0, efficiency, and error_tracker
• Missing (not important) p1_estimate, p2_estimate

A2. OE time-series (OE_deltaR1)

Inputs:

• IRE/T1.nii.gz and IRE/efficiency.nii.gz maps (from OE T1-mapping)
• OE/OE_dyn.nii.gz volume
• Parameters: oe_limits, average_fun, ...

Outputs:

• Maps (OE_output_maps/*.nii.gz) for R1, delta_R1, R1_baseline, R1_enhancing, R1_p_vals, and S_p_vals.

B1. DCE VFA T1-mapping (madym_T1 --T1_method VFA)

Inputs:

• VFA/VFA_*.nii.gz volumes (with *.json metadata)
• Parameters: T1_noise; TR, FA (read from metadata)

Outputs:

• Maps (T1_VFA/*.nii.gz) for T1, M0, and error_tracker

B2. DCE time-series, using first echo (DCE_deltaCt)

Inputs:

• T1_VFA/T1.nii.gz map (from VFA T1-mapping)
• DCE/dce_dyn_echo1.nii.gz volume
• Parameters: dce_limits, relax_coeff, average_fun, ...

Outputs:

• Maps (DCE_output_maps/*.nii.gz) for Ct, delta_C, C_baseline, C_enhancing, C_p_vals, and S_p_vals.

B3. (madym_DCE --model ETM)

Inputs:

• T1_VFA/T1.nii.gz and error_tracker maps (from VFA T1-mapping)
• DCE/dce_dyn_echo1.nii.gz volume
• Parameters: dose, hct, iauc, inj, ...

Outputs (??? - untested!)

• AUC 60 (yes, ETM/IAUC60.nii.gz)
• Binary map of AUC 60 > 0 (no)

[Combined] significance masks (OE_DCE_summary)

Inputs:

• *_p_vals maps from OE_output_maps and DCE_output_maps
• sig_level currently (hard-set?) at 5%
• ROI map (expected to be set semi-interactively)

Outputs:

• Significance masks for R1, S(t), and C(t) at sig_level
• Forman and Bonferroni corrected significance masks
• Combined masks

CHECK: are output map names *_5pct_* adjusted to sig_level?

Expected output maps:

OE IR T1-mapping

• Baseline T1 (yes, madym_output/T1_IR/T1.nii.gz)
• Baseline R1 (no)
• p1_estimate (?)
• p2_estimate (?)

OE time-series

• Delta R1(t) (yes, OE_output_maps/delta_R1.nii.gz)

• P-values for enhancing R1(t) (yes, OE_output_maps/R1_p_vals.nii.gz)

• P-values for enhancing S(t) (yes, OE_output_maps/S_p_vals.nii.gz)

• Significance mask for R1(t) at 5% (yes, OE_output_maps/R1_p_vals_sig_5pct.nii.gz)

• Significance mask for S(t) at 5% (yes, OE_output_maps/S_p_vals_sig_5pct.nii.gz)

• Significance mask for R1(t) at 5%, Bonferroni corrected (yes, OE_output_maps/R1_p_vals_sig_5pct_bf.nii.gz)

• Significance mask for S(t) at 5%, Bonferroni corrected (yes, OE_output_maps/S_p_vals_sig_5pct_bf.nii.gz)

• Significance mask for R1(t), Forman corrected (yes, OE_output_maps/R1_p_vals_sig_forman.nii.gz)

• Significance mask for S(t), Forman corrected (yes, OE_output_maps/S_p_vals_sig_forman.nii.gz)

DCE VFA T1-mapping

• Baseline T1 (yes, madym_output/T1_VFA/T1.nii.gz)
• Baseline R1 (no)

DCE time-series, using first echo

• Delta C(t) (yes, DCE_output_maps/delta_C.nii.gz)

• P-values for enhancing C(t) (yes, DCE_output_maps/C_p_vals.nii.gz)

• P-values for enhancing S(t) (yes, DCE_output_maps/S_p_vals.nii.gz)

• AUC 60 (yes, madym_output/ETM_pop/IAUC60.nii.gz)

• Binary map of AUC 60 > 0 (no)

• Significance mask for C(t) at 5% (yes, DCE_output_maps/C_p_vals_sig_5pct.nii.gz)

• Significance mask for S(t) at 5% (yes, DCE_output_maps/S_p_vals_sig_5pct.nii.gz)

Combined

• combinedimage_R1_5pc combinedmask_60_R1_5pc_
• combinedimage_S_5pc combinedmask_S_60_5pc_
• combinedimage_AUC60_5pc combinedmask_AUC60_5pc_