• In-vitro MRI datasets and mode...

Treffer: In-vitro MRI datasets and models used in the manuscript: 'Accelerated and quantitative three‐dimensional molecular MRI using a generative adversarial network'

Title:
In-vitro MRI datasets and models used in the manuscript: 'Accelerated and quantitative three‐dimensional molecular MRI using a generative adversarial network'
Authors:
Jonah Weigand-Whittier (13124550), Or Perlman (10844745), Christian T. Farrar (11046865), Moritz Zaiss (11046859), Christopher Nguyen (13124666), Elizabeth R. Gerstner (5395073), Anna N. Foster (13124670), Jauma Coll-Faunt (13124672), Kai Herz (13124674), Maria Sedykh (11569312)
Publication Year:
2022
Subject Terms:
Agricultural biotechnology not elsewhere classified, Animal physiology - biophysics, Radiology and organ imaging, Human biophysics, Neurosciences not elsewhere classified, Other biomedical and clinical sciences not elsewhere classified, Medical physics, Chemical exchange saturation transfer (CEST), Magnetization transfer (MT), Quantitative imaging, MRI, Magnetic resonance fingerprinting (MRF), Phantom, Generational adversarial network (GAN), Deep learning, Biological Engineering, Neuroscience, Medicine, Biophysics
Document Type:
E-Ressource software
Language:
unknown
Relation:
https://figshare.com/articles/software/In-vitro_MRI_datasets_and_models_used_in_the_manuscript_Accelerated_and_Quantitative_3D_Molecular_MRI_using_a_Generative_Adversarial_Network_/20346369
DOI:
10.6084/m9.figshare.20346369.v2
Availability:
https://doi.org/10.6084/m9.figshare.20346369.v2
Rights:
CC BY 4.0
Accession Number:
edsbas.DABB08CC
Database:
BASE

Weitere Informationen

GAN-ST Sample Models and Datasets The purpose of this code is to demonstrate CEST parameter mapping from 9 raw MRF images using GAN-ST. This folder contains the following: Data These data are comprised of single, reprasentative L-arinine phantom slices (6 vials with concentrations 25-100mM and pH between 4 and 6). Images were acquired using 3T clinical scanners. sample_phantom_input.npy contains the 9 raw MRF images used as input. sample_phantom_gt_ksw.npy contains ground truth CEST-MRF concentration and proton exchange rate parameter maps. sample_phantom_gt_ph.npy contains ground truth concentration and pH maps generated from pH-meter measured pH and analytic scale L-arginine concentrations. Brain and leg MRF input data and ground truth parameter maps are not included due to participant/patient privacy. Models The 'Models' folder contains 4 trained GAN-ST models. model_phantom_ksw.h5 maps 9 raw MRF images to concentration and proton exchange rate parameter maps for an L-arginine phantom. model_phantom_ph.h5 maps 9 raw MRF images to concentration and pH parameter maps based on gold-standard measured values for an L-arginine phantom. model_leg.h5 maps 9 raw MRF images to semi-solid volume fraction and exchange rate parameter maps for in-vivo calf images. model_brain.h5 maps 9 raw MRF images to semi-solid volume fraction and exchange rate parameter maps for in-vivo brain images. Python script sample_processing.py loads both in-vitro model files and maps a N = 9 MRF images to either concentration and proton exchange rate or concentration and pH maps. It requires having Python installed with the following packages: numpy, matplotlib, and keras. Suggested installation routes: pip Anaconda Once the packages are installed, run sample_processing.py The script will fit the models and output labeled CEST-MRF (ground truth) and GAN-ST parameter maps. These datasets are associated with the paper: Weigand‐Whittier J, Sedykh M, Herz K, et al. Accelerated and quantitative three‐dimensional molecular MRI using a generative ...