Physics-Informed autoencoder for DSC-MRI Perfusion post-processing: application to glioma grading

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Fayolle, Pierre, Bône, Alexandre, Debs, Noëlie, Naudin, Mathieu, Bourdon, Pascal, Guillevin, Remy, Helbert, David
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917018286948352
author Fayolle, Pierre
Bône, Alexandre
Debs, Noëlie
Naudin, Mathieu
Bourdon, Pascal
Guillevin, Remy
Helbert, David
author_facet Fayolle, Pierre
Bône, Alexandre
Debs, Noëlie
Naudin, Mathieu
Bourdon, Pascal
Guillevin, Remy
Helbert, David
contents DSC-MRI perfusion is a medical imaging technique for diagnosing and prognosing brain tumors and strokes. Its analysis relies on mathematical deconvolution, but noise or motion artifacts in a clinical environment can disrupt this process, leading to incorrect estimate of perfusion parameters. Although deep learning approaches have shown promising results, their calibration typically rely on third-party deconvolution algorithms to generate reference outputs and are bound to reproduce their limitations. To adress this problem, we propose a physics-informed autoencoder that leverages an analytical model to decode the perfusion parameters and guide the learning of the encoding network. This autoencoder is trained in a self-supervised fashion without any third-party software and its performance is evaluated on a database with glioma patients. Our method shows reliable results for glioma grading in accordance with other well-known deconvolution algorithms despite a lower computation time. It also achieved competitive performance even in the presence of high noise which is critical in a medical environment.
format Preprint
id arxiv_https___arxiv_org_abs_2510_13886
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Physics-Informed autoencoder for DSC-MRI Perfusion post-processing: application to glioma grading
Fayolle, Pierre
Bône, Alexandre
Debs, Noëlie
Naudin, Mathieu
Bourdon, Pascal
Guillevin, Remy
Helbert, David
Quantitative Methods
Artificial Intelligence
Image and Video Processing
Signal Processing
DSC-MRI perfusion is a medical imaging technique for diagnosing and prognosing brain tumors and strokes. Its analysis relies on mathematical deconvolution, but noise or motion artifacts in a clinical environment can disrupt this process, leading to incorrect estimate of perfusion parameters. Although deep learning approaches have shown promising results, their calibration typically rely on third-party deconvolution algorithms to generate reference outputs and are bound to reproduce their limitations. To adress this problem, we propose a physics-informed autoencoder that leverages an analytical model to decode the perfusion parameters and guide the learning of the encoding network. This autoencoder is trained in a self-supervised fashion without any third-party software and its performance is evaluated on a database with glioma patients. Our method shows reliable results for glioma grading in accordance with other well-known deconvolution algorithms despite a lower computation time. It also achieved competitive performance even in the presence of high noise which is critical in a medical environment.
title Physics-Informed autoencoder for DSC-MRI Perfusion post-processing: application to glioma grading
topic Quantitative Methods
Artificial Intelligence
Image and Video Processing
Signal Processing
url https://arxiv.org/abs/2510.13886