Sequenzadaptive Multiparametrische Magnetresonanzbildgebungsquantifizierung eines Organs mit Tiefenlernen
Anmelder: Universität Bern 🇨🇭
Details
- Veröffentlichungs-Nr.
- EP4732767
- Anmeldetag
- 22. Oktober 2024
- Veröffentlichung
- 29. April 2026
- Rechtsraum
- EP
- IPC
- A61B5/055, G06T7/11, G06T7/00(BAYER AG et al.)
Abstract
A computer-implemented method is proposed for quantifying one or more regions of interest (ROIs) of an organ or tumour. The method comprises: acquiring a first MRI sequence of a first sequence type, referred to a source sequence, and one or more second MRI sequences of at least a second sequence type, referred to as target sequences, the source and target sequences representing the ROIs of the organ or tumour of a patient under examination; segmenting the source sequence by using a trained AI model to obtain a labelled source sequence including a set of labels; transferring the labels from the source sequence to the target sequence(s) to obtain one or more labelled target sequences enabling identification of the ROIs in the labelled target sequence(s); and calculating ROI parametric information from voxel values identified by a respective label of the labelled target sequence(s) as processed or unprocessed to obtain region-specific information about the ROls.
Anmelder
- Firma
- Universität Bern
- Land
- 🇨🇭 Schweiz
2018 in Bern gegründete, international anerkannte Kanzlei für Patent- und Markenrecht, ansässig nahe dem Eidgenössischen Institut für Geistiges Eigentum. Berät Konzerne, Universitäten und Erfinder zu Schutzrechten, Prozessvertretung und Lizenzen, mit Schwerpunkten in Uhrenindustrie und Medizintechnik, dreisprachig.