Logo of the Physikalisch-Technische Bundesanstalt

AI-powered Method for estimation of biophysical tissue properties


The PTB Working Group 8.13 has developed a new AI-powered method to improve the reproducible measurement of biophysical parameters in tissue using Magnetic Resonance Imaging (MRI).

The so-called quantitative MRI (qMRI) faces the challenge of calculating the desired tissue parameter maps from the raw data of the MRI device. This is a complex and non-linear problem. While classical solutions are based on knowledge of the underlying physics of signal generation and hand-crafted assumptions about the distribution of measurement values, novel AI-powered methods learn the distribution of artifacts and tissue parameters based on data. However, previous AI methods have struggled to fully leverage the existing knowledge of underlying MR physics.

In their paper, Felix F. Zimmermann, Christoph Kolbitsch, Patrick Schuenke, and Andreas Kofler describe their new method called PINQI. This integrates knowledge of the signal model and learned regularization into a single trainable neural network. By using inner optimization blocks and learned network blocks, PINQI enables a reconstruction that is consistent with the acquired measurement data. PINQI can be trained with purely synthetic data and delivers superior results not only in synthetic benchmarks but also when applied to in vivo data acquired with the PTB's own MRI scanner. 

The results were published in the renowned journal "IEEE Transactions on Computational Imaging"  (DOI 10.1109/TCI.2024.3388869).


Contact: Christoph Kolbitsch


Head of Press and Information Office

Dr. Dr. Jens Simon

Phone: +49 531 592-3005


Physikalisch-Technische Bundesanstalt
Bundesallee 100
38116 Braunschweig