Verwendung eines Neuronalen Netzwerks zur Automatischen Rekonstruktion eines Dreidimensionalen Knochenmodells aus Mrt-Scan-Daten und Verfahren zum Training davon

Details

Veröffentlichungs-Nr.

EP4797204

Anmeldetag

24. Februar 2025

Veröffentlichung

26. August 2026

Rechtsraum

EP

IPC

G06T7/11

Abstract

A method is provided herein for training a neural network to automatically reconstruct a three-dimensional bone model from MRI-scan data. The method that is performed for a plurality of human body part samples comprises preparing (S1M, S1C) a respective pair of an MRI-scan comprising MRI-scan data (M) and a CT-scan with CT-scan data (C). Ground truth bone segmentation data (Cs) is provided (S2C) for one or more regions of interest in the CT-scan. The ground truth bone segmentation data (Cs) is linked (S4C) with the CT-scan data to obtain enriched CT-scan data (Cl). The MRI-scan data is fused (S5) with the enriched CT-scan data (Cl) to obtain enriched MRI-scan data (Fa, Fb) including ground truth bone segmentation data. Then with the neural network to be trained tentative enriched MRI-scan data (Ft) is estimated (S7) from the MRI-scan data. The tentative enriched MRI-scan data (Ft) comprises the MRI-scan data and tentative bone segmentation data. A loss is computed(S8) that is indicative for a deviation between the tentative bone segmentation data and the ground truth bone segmentation data and backpropagated (S9). The trained neural network (20) obtained with the method is suitable for computing bone segmentation data from MRI-scan data.

Anmelder

Firmen

RIJKSUNIVERSITEIT GRONINGEN
Academisch Ziekenhuis Groningen

Land

🇳🇱 Niederlande

Fachgebiete