p-Brain: An Automated MRI Pipeline for Cerebral Perfusion, Microvasculature, and Blood-Brain Barrier Permeability Estimation

Abstract

We present p-Brain, an end-to-end neuroimaging analysis framework for reproducible, automated quantitative DCE-MRI analysis at scale. From standard acquisitions, p-Brain estimates baseline relaxation parameters, converts signal to gadolinium concentration, derives arterial and venous input functions using convolutional neural network (CNN) slice selection and ROI segmentation, and produces voxelwise maps with regional and whole-brain summaries. The pipeline implements Patlak graphical analysis to estimate the blood-brain barrier influx constant (Ki) and plasma volume fraction (vp), and performs model-free residue deconvolution with Tikhonov regularisation to estimate cerebral blood flow (CBF), mean transit time (MTT), and capillary transit-time heterogeneity (CTH) from the same DCE dataset. p-Brain exports analysis-ready outputs, intermediate readouts, structured runtime metadata, and stage-level quality control artifacts to support auditability in batch processing. We evaluate the framework on a technically uniform set of 97 DCE-MRI scans from 58 healthy human participants, and show close agreement between automated Patlak Ki summaries and an established reference workflow. A companion macOS desktop application supports batch execution, job monitoring, and rapid review of curves and maps. p-Brain is open-source and configurable, enabling extension to additional kinetic models.