Introduction of R&D Environments for Synthetic Q-space Learning : An Instance

To evaluate recent research and development in machine learning and artificial intelligence (AI) applications not limited to MRI research, it is necessary to consider various factors, including the choice of operating system, programming language, application programming interface, machine learning frameworks, and use of parallel computing and batch processing for faster and more efficient computation. Each option offers certain advantages and disadvantages, and it must be assumed that researchers, including myself, are constantly engaged in a trial-and-error process, often without knowing whether the environments are optimal. This article introduces an instance of research on synthetic Q-space learning (synQSL), a method for robust and fast estimation of the parameters of signal value models in diffusion MRI (dMRI), which we are intensively working on. In addition to the research environment used (e.g., equipment, operating system, and software), this paper describes the actual situation of the field, including the efficiency and acceleration of the experimental process, release of freeware, and porting of software to high-performance computing environments. We hope that this case study will serve as a reference for the researchers working in related fields.