3D-computed tomography (CT) reconstruction is the task of generating volumetric images from 2D X-ray images. X-ray beams, when passed through a body part, generate a single 2D slice, and the next slice of the X-ray image can be generated by slightly changing the position of the patient. Such 2D X-ray images are unable to capture a detailed volumetric view of the body part. The use of 3D scanners is harmful, since they expose the patient for a longer duration to X-ray beams as well as being highly costly. This motivates the design of 3D volumetric reconstruction algorithms that can generate a 3D view of the object from 2D X-ray images. Recent advances in artificial intelligence and deep learning have provided a great impetus to create volumetric images of 2D X-rays. This has greatly enabled a better understanding of diseases through taking a volumetric view of the body part. CT reconstruction from 2D X-ray images can be categorized into three techniques. Firstly, by using a single X-ray image, a deep learning model can estimate multiple 2D slices of various depths, thereby giving a 3D understanding. Secondly, given multiple X-ray images as input, a deep learning model can generate even more X-ray images, thereby enabling further insight. Lastly, by using one or more X-ray images, a deep learning model can generate a 3D structure of the body part scanned. In this chapter, we will focus on the imaging techniques and principles of computed tomography, their main drawbacks, and the motivation behind deep learning-based 3D reconstruction. Then, we will discuss in detail the three techniques of volumetric reconstruction from two-dimensional X-ray images.