Deep neural network approaches for dictionary free data analysis in magnetic resonance fingerprinting.
Barbieri M., Brizi L.
V - Biofisica e fisica medica
Aula 32C-2 - Mercoledì 19 h 09:00 - 13:00
» Download abstract
The feasibility of using Neural Network approaches to Magnetic Resonance Fingerprinting (MRF) is investigated for overcoming the limitations of the original methodology, based on the matching of the MRF data with precomputed dictionaries, which are inefficiencies in memory usage and computational time when large dictionaries are used. The proposed method faces these issues taking advantage of Machine Learning algorithms for the prediction of the MR parameters given the experimental fingerprint as the input. The results indicate that the approach is computationally efficient in predicting MR parameters with accuracy and noise robustness.