Machine learning (ML) has been increasingly used to quantify chemical exchange saturation transfer (CEST) effect. ML models are typically trained using either measured data or fully simulated data. However, training with measured data often lacks sufficient training data, while training with fully simulated data may introduce bias due to limited simulations pools. This study introduces a new platform that combines simulated and measured components to generate partially synthetic CEST data, and to evaluate its feasibility for training ML models to predict amide proton transfer (APT) effect. Partially synthetic CEST signals were created using an inverse summation of APT effects from simulations and the other components from measurements. Training data were generated by varying APT simulation parameters and applying scaling factors to adjust the measured components, achieving a balance between simulation flexibility and fidelity. First, tissue-mimicking CEST signals along with ground truth information were created using multiple-pool model simulations to validate this method. Second, an ML model was trained individually on partially synthetic data, in vivo data, and fully simulated data, to predict APT effect in rat brains bearing 9L tumors. Experiments on tissue-mimicking data suggest that the ML method using the partially synthetic data is accurate in predicting APT. In vivo experiments suggest that our method provides more accurate and robust prediction than the training using in vivo data and fully synthetic data. Partially synthetic CEST data can address the challenges in conventional ML methods.

While AI designers are creating powerful machines that can beat humans at many complex cognitive endeavors, they're still envious of the human brain's facility with certain seemingly simple tasks: such as recognizing a face after seeing it only once or when it's partially obscured. The U.S. intelligence agency IARPA is particularly interested in developing AI deep learning programs with visual recognition skills, so last year it launched a US $100 million program called Microns. Under Microns, three teams of researchers are looking for answers on the micro scale and in rodent brains. Each team is studying 1 cubic millimeter of brain tissue from a rodent's visual cortex, using precision instruments to map the 50,000 neurons and 500 million neural connections within that chunk. The researchers hope to discover patterns of neural activation that can be translated to architectures for AI programs known as deep neural networks.

A partial digital reconstruction of the brain previously made by Harvard. We've talked a lot about making a computer that works like the mammalian brain. The U.S. government is now betting $28 million dollars that all these projects are wrong. A series of three grants snagged by Harvard University from Intelligence Advanced Research Projects Activity (IARPA) last week has funded a "moonshot" project to throw out all of the previous attempts at understanding the brain and start fresh. But while DARPA focuses on military projects, IARPA focuses on intelligence agency research.)