Organs at risk (OARs) are healthy tissues around cancers that must be preserved in radiotherapy (RT). Heart is one of the fundamental organs for the full functioning of the human body, protecting this organ in the RT is of paramount importance. For this, the planning process must be careful. Planning begins with manual segmentation by specialists in computed tomography (CT). We propose a deep learning method for heart segmentation from planning CT. The method consists of four steps: (1) material acquisition from a public database; (2) volume standardisation using registration and histogram matching; (3) Volume of Interest (VOI) segmentation using atlas; and (4) a U-Net variation with 2.5D input layer for final segmentation. We tested in 36 CTs and achieved 93.82% of the Dice similarity coefficient, 88.42% of the Jaccard. With the innovation of the proposed method and the promising results, we show that our method effectively uses heart segmentation.