The present technology offers a machine-learning solution to the problem of fitting multi-exponential models to observed data. The multi-exponential fitting problem appears in various science and engineering applications, such as nuclear magnetic resonance spectroscopy, lattice quantum chromodynamics, pharmaceutics and chemical engineering, fluorescence imaging, infra-red imaging, economic model prediction, medical imaging, and more.
MELoDee (Multi-Exponential model Learning based on Deep neural networks), is an innovative deep-learning-based solver with improved generalization capabilities compared to previous solutions. The MELoDee consists of a new neural network architecture as well as a novel training protocol, aimed at producing a multi-exponential fitting solution that is more accurate, precise, and robust to uncertainties in the training conditions. The main innovations of this method are (a) a novel architecture, as well as (b) a new training protocol. These two features enable the network to generalize the multi-exponential model better than existing approaches, and thus yield more accurate and robust model parameter estimations.
- Fast, accurate and precise parameter estimation
- Robustness to uncertainties
Applications and Opportunities
- In-vivo non-invasive tissue characterization with quantitative MRI
- Cardiac inflammation mapping
- Brain microstructure assessment for neurodegenerative diseases
- Cancer imaging
- Accelerated MRI reconstruction
- Numerical software for signal analysis