Assigning jobs that typically a human expert designs based on his/her domain knowledge often in an analytic form or in an iterative form to the computing resources is facing a new paradigm. While such a highly intelligent and insightful approach by human experts usually works well with a problem that exploits single-data, deep learning based approaches can much better address problems that are supported by the so-called big-data. With the advancement in machine learning technologies, a wide spectrum of applications are explosively being developed. We have used a deep neural network for synthesizing unmeasured sinogram data in a sparse-view CT imaging application (IEEE Tran. Rad. Plas. Med. Sci. 2019) and are actively investigating other techniques based on deep learning. Treatment planning research based on deep learning in RT is also under our interest (Sci. Rep. in review, 2019).