Detailed Pipeline Example: Kinetic Modeling with SRTM, Brain VOIs in PET Space

Available Data

The pipeline discussed in detail below assumes the following available scenario: A patient had two separate scanning sessions, resulting in a dynamic brain PET and a non-matched T₁-weighted MRI.

Intended Analysis

The MRI, which has superior resolution and anatomical information, is used for the automatic outlining of brain regions. These VOIs are then projected to the PET, and the average PET uptake in each region is calculated. The resulting TACs are analyzed with the SRTM kinetic model using the average cerebellum uptake as the reference TAC.

Processing Pipeline Outline

It is assumed that the user has imported the data into a PMOD database and associated the PET and the MR series. The described analysis can then be implemented by the following pipeline sequence.

1. Because the anatomical information in the MR image is better than in the PET, the MR is first loaded into the pipeline. So the MR is the Input study.
2. The Automatic Brain VOIs tool is then applied. It generates the VOIs in the MR image space. These VOIs will continue existing in the following pipeline stages.
3. The rigid transformation between the MR and the PET is determined using the Registration & Normalization tool. Because the uptake calculation should use the non-interpolated PET values the PET image serves as the reference image, so that the rigid transform goes from the MR to the PET space. This rigid transform will continue existing in the following pipeline stages.
4. The PET images are brought into the pipeline using the Replace Images tool which swaps the current (MR) image with its associate (PET). From now on, the associated image is the MR.
5. If necessary, patient motion in the dynamic brain PET can be corrected with the Motion Correction tool. Note however that after this step the PET values will be interpolated.
6. Transform the VOIs from the MR to the PET space using the result from 3. (still in the pipeline), calculate the average regional TACs, generate the reference TAC by merging all cerebellum TACs, and perform the SRTM kinetic modeling. All this is done by the Kinetic Modeling tool.

In practice, the pipeline has the following appearance. Note that this pipeline setup is only possible if the Kinetic Modeling tool (PKIN) and the Image Fusion tool (PFUS) have been licensed from PMOD.

The following sections describe each tool configuration in detail. The steps can be reproduced with the PKIN2 data set in the Pmod example database. Furthermore, the corresponding pipeline definition is already available and can be retrieved with Load Processing.

Required PMOD Modules

Note that this pipeline requires that the following modules are available in the PMOD license:

• Image fusion (PFUS): for using the Automatic Brain VOIs, Registration & Normalization, Motion Correction tools.
• Neuro tool (PNEURO): for using the Hammers atlas. However, the AAL atlas can be used alternatively if PNEURO is not available.
• Kinetic modeling (PKIN): for applying the SRTM model to the tissue TACs.