Chemical Exchange Saturation Transfer (CEST) MRI uses selective radio-frequency pulses to saturate the magnetization of exchangeable protons on a variety of molecules, including proteins and metabolites, which due to fast chemical exchange with bulk water results in a decreased water MRI signal. The CEST image contrast depends on the chemical exchange rate, which is pH sensitive, and the volume fraction of the exchangeable proton pool, which is sensitive to protein and metabolite concentrations. The sensitivity of CEST MRI to pH and protein/metabolite concentrations has proven to be a powerful tool for imaging a wide range of disease pathologies. For example, the amide proton CEST contrast from endogenous proteins has recently been used to distinguish tumor recurrence from radiation necrosis, detect early tumor response to temozolomide and radiation therapies, evaluate the tumor grade and cellularity of clinical glioma patients, and detect changes in pH during stroke that may provide insight into the viability of the ischemic penumbra. However, clinical translation of these CEST-MRI methods has been hindered by the qualitative nature of the CEST contrast, the long image acquisition times, and the complex data processing required. Clinical translation of CEST MRI would benefit greatly from the development of more specific, quantitative and rapid CEST methods.
We have recently demonstrated the first use of a fast CEST fingerprinting method for generating quantitative exchange rate and exchangeable proton concentration maps of L-Arginine phantoms with different concentrations (25-100 mM) and pH (pH 4-6), in vivo rat brain stroke models, and in vivo mouse brain tumor models. Most recently we have demonstrated clinical translation of the CEST-MRF method in a normal human brain. The MRF method varies the image acquisition parameters to generate unique signal trajectories for different quantitative tissue parameters. The experimental trajectories are then pattern-matched to a large dictionary of signal trajectories simulated using the Bloch-McConnell equations for different combinations of tissue parameters. The MRF method allows for the simultaneous quantification of multiple parameters in a short acquisition time period.
Dictionary based methods for MRF parameter map reconstruction are, however, typically limited to matching only a few parameters since the size of the dictionary grows exponentially with the number of parameters making the matching of highly multi-parametric data computationally intensive and time consuming. In addition, the accuracy of the parameter maps is limited by the resolution of the matching dictionary, which is simulated for discrete parameter values only. The application of MRF methods for many disease pathologies, in which multiple tissue parameters are changing with time, is therefore very challenging, as a wide range of values for multiple tissue parameters will need to be included in the matching dictionary. Large dictionaries result in long reconstruction times, but also require significant computational resources to generate in the first place. To overcome these challenges, we have recently implemented a Deep Learning Reconstruction Neural Network (DRONE) based MRF reconstruction method that trains a neural network (NN) with sparse dictionaries. The neural network approach has the advantage of being both more efficient and accurate as the NN generates continuous parameter values instead of being limited to discrete values.