With an annual incidence of 18.8 cases per 100,000 and survival rates ranging from months to a few years, bone is the third most prevalent location of metastasis in the human body across all malignancies. Bone metastases most often originate from haematologic, breast, prostate, and lung cancers. Rare primary bone sarcomas have a better survival rate and an annual incidence of 0.9 per 100,000 people. Segmenting bones in abdominal CT images using a deep learning method. When working with medical pictures, segmentation is a typical first step that is essential for computer-aided detection and diagnostic systems. Currently, there is no widely accepted automated method for the difficult and time-consuming procedure of bone extraction from CT images, even when done manually by specialists. An end-to-end trained convolutional neural network that executes semantic data segmentation forms the basis of the offered approach; this network draws inspiration from the U-Net. Thirteen CT scans of the abdomen (ranging in size from 403 to 994 2D transversal images apiece) make up the training dataset. Each voxel in these high-resolution, 512x512 voxel pictures is labelled by the network as either "background," "femoral bones," "hips," "sacrum," "sternum," "spine," or "rib. Consequently, a bone mask with identified and classified bones is the end result.