Spine diseases are very common in all ages of humans due to lifestyle changes. Accurate spine segmentation is very crucial for detecting various spine diseases like fractures, herniated discs, and spinal stenosis, which involves advanced imaging techniques. This paper focuses on segmenting Cervical fractures from spine computed tomography (CT) Images. Deep learning model, U-Net, is applied for the segmentation on a dataset of 1075 CT scan images. The performance of the U-Net model is evaluated in terms of Dice score and Intersection over Union (IoU) performance metrics. The proposed U-Net Convoluted Neural Network (UNCNN) achieves dice score of 89.38% and an IoU of 81.24%. A comparative study is given that shows our model gives better result for segmentation. Training loss and validation loss are showing good results with the increase in epochs up to 30. .