One of the indicators of possible occurrences of cardiovascular diseases is the amount of coronary artery calcium. Recently, approaches using new technologies such as deep learning have been used to help identify these indicators. This work proposes a segmentation method for calcification of the coronary arteries that has 3 steps: extraction of the ROI using U-Net with batch normalization after convolution layers, segmentation of the calcifications and removal of false positives using U-Net with EfficientNetB0. The method uses histogram matching as preprocessing in order to increase the contrast between tissue and calcification and normalize the different types of exams. Tests were performed between different architectures where the best approach achieved 96.9% F1-Score, 98.5% precision and 97,2\%recall in the OrcaScore Dataset.