The fully-automated PIANO™ pipeline for both exploratory voxelwise analysis as well as quantitative regional DMRI analysis was used to investigate the CBD and healthy control populations.
This illustration provides a high level overview of the various steps involved in our PIANO™ diffusion MRI (DMRI) processing "pipeline". Briefly the DMRI scans, including at least one b0 image and diffusion-weighted images acquired with multiple gradient encodings, are processed with the following steps:
- Motion correction utilizing a 12-parameter affine registration process.
- PIANO™ then fits the Diffusion Tensor Imaging (DTI) and Free Water models, generating fractional anisotropy (FA), free water (FW), and mean, axial, & radial diffusivity (MD, AD, RD) parameter maps.
- PIANO™ utilizes a proprietary, automated process that aligns each subject’s DMRI scan with that subject’s registration target’s anatomical MRI scan using nonlinear registration.
- Regional averages of DMRI metrics are then generated on pre-defined ROIs. Specific ROIs are defined anatomically, explicitly on a template, or generated implicitly using a deep-learning model (hippocampus), and are combined with subject-specific tissue classification, as needed.