Much of the recently available research and challenge data lack the meta-data containing any information about the patient position. This paper presents a tool for automatic rotation of CT data into a standardized (HFS) patient position. The proposed method is based on the prediction of rotation angle with CNN, and it achieved nearly perfect results with an accuracy of 99.55 %. We provide implementations with easy to use an example for both Matlab and Python (PyTorch), which can be used, for example, for automatic rotation correction of VerSe2020 challenge data.