Related Experiment Video
Updated: Jul 13, 2026
08:48
Neuronavigation-guided Repetitive Transcranial Magnetic Stimulation for Aphasia
Published on: May 6, 2016
12.4K
Neural networks with personalized training for improved MOLLI T1 mapping
Olympia Gkatsoni1, Christos G Xanthis2, Sebastian Johansson2
1Laboratory of Computing, Medical Informatics and Biomedical - Imaging Technologies, School of Medicine, Aristotle University of Thessaloniki, Thessaloniki, Greece.
BMC Medical Imaging
|July 2, 2025
Summary
Personalized Training Neural Network (PTNN) improves T1 mapping accuracy using MRI simulations. This novel method offers more precise T1 estimates in phantoms and volunteers compared to conventional techniques.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Cardiovascular MRI
Background:
- Accurate T1 mapping is crucial for cardiovascular assessment using Magnetic Resonance Imaging (MRI).
- Conventional T1 estimation methods can be limited by fitting inaccuracies and physiological variability.
- Deep Neural Networks (DNNs) offer potential for improving quantitative MRI analyses.
Purpose of the Study:
- To develop and validate a personalized deep neural network (PTNN) for improved T1 mapping using MRI simulation.
- To enhance the accuracy of MOLLI (Multi-Organ, Low-Inversion) T1 estimates compared to traditional fitting methods.
- To investigate the performance of PTNN across phantom and in vivo datasets.
Main Methods:
- A neural network was trained using simulated MOLLI signals tailored to individual scan parameters and heart rate triggers.
- The Personalized Training Neural Network (PTNN) approach was applied to T1 mapping.
- Data from eleven phantoms and ten healthy volunteers were utilized for validation.
Main Results:
- PTNN demonstrated significantly smaller bias in T1 estimates compared to conventional fitting in phantom studies.
- In vivo studies showed PTNN yielding higher T1 values for myocardium and blood compared to conventional fitting.
- Acquisition time reduction with PTNN (eliminating pause) still resulted in higher myocardial T1 values.
Conclusions:
- PTNN offers a post-processing method for generating T1 maps with enhanced accuracy and higher values than conventional fitting.
- The method performs well across a physiological range of T1 and T2 values in phantoms.
- PTNN achieves improved T1 estimates in volunteers, even with accelerated imaging protocols, without new pulse sequences.
